diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9492 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.995008319467554,
+  "eval_steps": 500,
+  "global_step": 1350,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0022185246810870773,
+      "grad_norm": 2.1247129440307617,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 9.699,
+      "step": 1
+    },
+    {
+      "epoch": 0.004437049362174155,
+      "grad_norm": 2.200881004333496,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 9.5438,
+      "step": 2
+    },
+    {
+      "epoch": 0.0066555740432612314,
+      "grad_norm": 2.15293288230896,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 9.5488,
+      "step": 3
+    },
+    {
+      "epoch": 0.00887409872434831,
+      "grad_norm": 2.195706844329834,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 9.4276,
+      "step": 4
+    },
+    {
+      "epoch": 0.011092623405435386,
+      "grad_norm": 2.062354803085327,
+      "learning_rate": 1.111111111111111e-05,
+      "loss": 9.4862,
+      "step": 5
+    },
+    {
+      "epoch": 0.013311148086522463,
+      "grad_norm": 2.240753650665283,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 9.4234,
+      "step": 6
+    },
+    {
+      "epoch": 0.01552967276760954,
+      "grad_norm": 2.088193416595459,
+      "learning_rate": 1.5555555555555555e-05,
+      "loss": 9.4802,
+      "step": 7
+    },
+    {
+      "epoch": 0.01774819744869662,
+      "grad_norm": 2.2427570819854736,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 9.7638,
+      "step": 8
+    },
+    {
+      "epoch": 0.019966722129783693,
+      "grad_norm": 2.54486346244812,
+      "learning_rate": 1.9999999999999998e-05,
+      "loss": 9.2772,
+      "step": 9
+    },
+    {
+      "epoch": 0.022185246810870772,
+      "grad_norm": 2.3003125190734863,
+      "learning_rate": 2.222222222222222e-05,
+      "loss": 9.6489,
+      "step": 10
+    },
+    {
+      "epoch": 0.024403771491957847,
+      "grad_norm": 2.762918710708618,
+      "learning_rate": 2.4444444444444445e-05,
+      "loss": 9.3257,
+      "step": 11
+    },
+    {
+      "epoch": 0.026622296173044926,
+      "grad_norm": 2.6459238529205322,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 9.3666,
+      "step": 12
+    },
+    {
+      "epoch": 0.028840820854132,
+      "grad_norm": 2.8060708045959473,
+      "learning_rate": 2.8888888888888885e-05,
+      "loss": 9.4591,
+      "step": 13
+    },
+    {
+      "epoch": 0.03105934553521908,
+      "grad_norm": 2.7448670864105225,
+      "learning_rate": 3.111111111111111e-05,
+      "loss": 9.3005,
+      "step": 14
+    },
+    {
+      "epoch": 0.033277870216306155,
+      "grad_norm": 2.9884493350982666,
+      "learning_rate": 3.333333333333333e-05,
+      "loss": 9.413,
+      "step": 15
+    },
+    {
+      "epoch": 0.03549639489739324,
+      "grad_norm": 2.796663522720337,
+      "learning_rate": 3.555555555555555e-05,
+      "loss": 9.3922,
+      "step": 16
+    },
+    {
+      "epoch": 0.03771491957848031,
+      "grad_norm": 3.1795897483825684,
+      "learning_rate": 3.777777777777777e-05,
+      "loss": 9.111,
+      "step": 17
+    },
+    {
+      "epoch": 0.03993344425956739,
+      "grad_norm": 3.041274070739746,
+      "learning_rate": 3.9999999999999996e-05,
+      "loss": 9.0777,
+      "step": 18
+    },
+    {
+      "epoch": 0.04215196894065446,
+      "grad_norm": 3.1814870834350586,
+      "learning_rate": 4.222222222222222e-05,
+      "loss": 9.0334,
+      "step": 19
+    },
+    {
+      "epoch": 0.044370493621741544,
+      "grad_norm": 3.389146566390991,
+      "learning_rate": 4.444444444444444e-05,
+      "loss": 8.8284,
+      "step": 20
+    },
+    {
+      "epoch": 0.04658901830282862,
+      "grad_norm": 3.4111945629119873,
+      "learning_rate": 4.6666666666666665e-05,
+      "loss": 8.6863,
+      "step": 21
+    },
+    {
+      "epoch": 0.048807542983915694,
+      "grad_norm": 3.2762200832366943,
+      "learning_rate": 4.888888888888889e-05,
+      "loss": 8.4764,
+      "step": 22
+    },
+    {
+      "epoch": 0.051026067665002776,
+      "grad_norm": 3.4036614894866943,
+      "learning_rate": 5.111111111111111e-05,
+      "loss": 8.7689,
+      "step": 23
+    },
+    {
+      "epoch": 0.05324459234608985,
+      "grad_norm": 3.7457964420318604,
+      "learning_rate": 5.333333333333333e-05,
+      "loss": 8.7579,
+      "step": 24
+    },
+    {
+      "epoch": 0.05546311702717693,
+      "grad_norm": 3.8012592792510986,
+      "learning_rate": 5.5555555555555545e-05,
+      "loss": 8.1657,
+      "step": 25
+    },
+    {
+      "epoch": 0.057681641708264,
+      "grad_norm": 3.8158586025238037,
+      "learning_rate": 5.777777777777777e-05,
+      "loss": 8.2045,
+      "step": 26
+    },
+    {
+      "epoch": 0.059900166389351084,
+      "grad_norm": 5.272186279296875,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 8.2431,
+      "step": 27
+    },
+    {
+      "epoch": 0.06211869107043816,
+      "grad_norm": 5.857285022735596,
+      "learning_rate": 6.222222222222222e-05,
+      "loss": 8.2881,
+      "step": 28
+    },
+    {
+      "epoch": 0.06433721575152523,
+      "grad_norm": 5.007485866546631,
+      "learning_rate": 6.444444444444444e-05,
+      "loss": 8.0562,
+      "step": 29
+    },
+    {
+      "epoch": 0.06655574043261231,
+      "grad_norm": 5.025998592376709,
+      "learning_rate": 6.666666666666666e-05,
+      "loss": 7.8378,
+      "step": 30
+    },
+    {
+      "epoch": 0.06877426511369938,
+      "grad_norm": 4.809839248657227,
+      "learning_rate": 6.888888888888888e-05,
+      "loss": 7.7989,
+      "step": 31
+    },
+    {
+      "epoch": 0.07099278979478647,
+      "grad_norm": 3.9237091541290283,
+      "learning_rate": 7.11111111111111e-05,
+      "loss": 7.5362,
+      "step": 32
+    },
+    {
+      "epoch": 0.07321131447587355,
+      "grad_norm": 3.669914960861206,
+      "learning_rate": 7.333333333333332e-05,
+      "loss": 7.5813,
+      "step": 33
+    },
+    {
+      "epoch": 0.07542983915696062,
+      "grad_norm": 3.283878803253174,
+      "learning_rate": 7.555555555555554e-05,
+      "loss": 7.3411,
+      "step": 34
+    },
+    {
+      "epoch": 0.0776483638380477,
+      "grad_norm": 3.651585817337036,
+      "learning_rate": 7.777777777777777e-05,
+      "loss": 7.0364,
+      "step": 35
+    },
+    {
+      "epoch": 0.07986688851913477,
+      "grad_norm": 3.2250444889068604,
+      "learning_rate": 7.999999999999999e-05,
+      "loss": 7.3274,
+      "step": 36
+    },
+    {
+      "epoch": 0.08208541320022185,
+      "grad_norm": 3.6796436309814453,
+      "learning_rate": 8.222222222222222e-05,
+      "loss": 7.4356,
+      "step": 37
+    },
+    {
+      "epoch": 0.08430393788130892,
+      "grad_norm": 4.444842338562012,
+      "learning_rate": 8.444444444444444e-05,
+      "loss": 7.0472,
+      "step": 38
+    },
+    {
+      "epoch": 0.08652246256239601,
+      "grad_norm": 4.753952503204346,
+      "learning_rate": 8.666666666666665e-05,
+      "loss": 6.7329,
+      "step": 39
+    },
+    {
+      "epoch": 0.08874098724348309,
+      "grad_norm": 3.6671204566955566,
+      "learning_rate": 8.888888888888888e-05,
+      "loss": 6.8531,
+      "step": 40
+    },
+    {
+      "epoch": 0.09095951192457016,
+      "grad_norm": 2.9891927242279053,
+      "learning_rate": 9.11111111111111e-05,
+      "loss": 7.499,
+      "step": 41
+    },
+    {
+      "epoch": 0.09317803660565724,
+      "grad_norm": 2.8181629180908203,
+      "learning_rate": 9.333333333333333e-05,
+      "loss": 6.6919,
+      "step": 42
+    },
+    {
+      "epoch": 0.09539656128674431,
+      "grad_norm": 2.4321961402893066,
+      "learning_rate": 9.555555555555555e-05,
+      "loss": 6.7912,
+      "step": 43
+    },
+    {
+      "epoch": 0.09761508596783139,
+      "grad_norm": 2.447136640548706,
+      "learning_rate": 9.777777777777778e-05,
+      "loss": 6.7424,
+      "step": 44
+    },
+    {
+      "epoch": 0.09983361064891846,
+      "grad_norm": 2.347606897354126,
+      "learning_rate": 9.999999999999999e-05,
+      "loss": 6.5965,
+      "step": 45
+    },
+    {
+      "epoch": 0.10205213533000555,
+      "grad_norm": 3.112799644470215,
+      "learning_rate": 0.00010222222222222222,
+      "loss": 6.9193,
+      "step": 46
+    },
+    {
+      "epoch": 0.10427066001109263,
+      "grad_norm": 2.9997987747192383,
+      "learning_rate": 0.00010444444444444444,
+      "loss": 6.6834,
+      "step": 47
+    },
+    {
+      "epoch": 0.1064891846921797,
+      "grad_norm": 2.2601795196533203,
+      "learning_rate": 0.00010666666666666667,
+      "loss": 6.4838,
+      "step": 48
+    },
+    {
+      "epoch": 0.10870770937326678,
+      "grad_norm": 2.697308301925659,
+      "learning_rate": 0.00010888888888888889,
+      "loss": 6.7889,
+      "step": 49
+    },
+    {
+      "epoch": 0.11092623405435385,
+      "grad_norm": 1.8948653936386108,
+      "learning_rate": 0.00011111111111111109,
+      "loss": 6.5378,
+      "step": 50
+    },
+    {
+      "epoch": 0.11314475873544093,
+      "grad_norm": 2.4612224102020264,
+      "learning_rate": 0.00011333333333333331,
+      "loss": 6.4858,
+      "step": 51
+    },
+    {
+      "epoch": 0.115363283416528,
+      "grad_norm": 1.8843343257904053,
+      "learning_rate": 0.00011555555555555554,
+      "loss": 6.4761,
+      "step": 52
+    },
+    {
+      "epoch": 0.11758180809761509,
+      "grad_norm": 1.881941795349121,
+      "learning_rate": 0.00011777777777777776,
+      "loss": 6.6741,
+      "step": 53
+    },
+    {
+      "epoch": 0.11980033277870217,
+      "grad_norm": 1.630995750427246,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 6.5392,
+      "step": 54
+    },
+    {
+      "epoch": 0.12201885745978924,
+      "grad_norm": 1.6674602031707764,
+      "learning_rate": 0.00012222222222222221,
+      "loss": 6.3921,
+      "step": 55
+    },
+    {
+      "epoch": 0.12423738214087632,
+      "grad_norm": 1.7581100463867188,
+      "learning_rate": 0.00012444444444444444,
+      "loss": 6.3696,
+      "step": 56
+    },
+    {
+      "epoch": 0.1264559068219634,
+      "grad_norm": 2.0795226097106934,
+      "learning_rate": 0.00012666666666666666,
+      "loss": 6.4294,
+      "step": 57
+    },
+    {
+      "epoch": 0.12867443150305047,
+      "grad_norm": 1.7071822881698608,
+      "learning_rate": 0.0001288888888888889,
+      "loss": 6.3776,
+      "step": 58
+    },
+    {
+      "epoch": 0.13089295618413754,
+      "grad_norm": 2.0405585765838623,
+      "learning_rate": 0.00013111111111111111,
+      "loss": 6.2884,
+      "step": 59
+    },
+    {
+      "epoch": 0.13311148086522462,
+      "grad_norm": 1.8677833080291748,
+      "learning_rate": 0.0001333333333333333,
+      "loss": 6.1985,
+      "step": 60
+    },
+    {
+      "epoch": 0.1353300055463117,
+      "grad_norm": 1.7754088640213013,
+      "learning_rate": 0.00013555555555555554,
+      "loss": 6.4073,
+      "step": 61
+    },
+    {
+      "epoch": 0.13754853022739877,
+      "grad_norm": 1.7457290887832642,
+      "learning_rate": 0.00013777777777777776,
+      "loss": 6.4382,
+      "step": 62
+    },
+    {
+      "epoch": 0.13976705490848584,
+      "grad_norm": 1.5886279344558716,
+      "learning_rate": 0.00014,
+      "loss": 6.4177,
+      "step": 63
+    },
+    {
+      "epoch": 0.14198557958957295,
+      "grad_norm": 2.2759571075439453,
+      "learning_rate": 0.0001422222222222222,
+      "loss": 6.3328,
+      "step": 64
+    },
+    {
+      "epoch": 0.14420410427066002,
+      "grad_norm": 1.9168124198913574,
+      "learning_rate": 0.0001444444444444444,
+      "loss": 6.3291,
+      "step": 65
+    },
+    {
+      "epoch": 0.1464226289517471,
+      "grad_norm": 1.8469194173812866,
+      "learning_rate": 0.00014666666666666664,
+      "loss": 6.4009,
+      "step": 66
+    },
+    {
+      "epoch": 0.14864115363283417,
+      "grad_norm": 1.7399849891662598,
+      "learning_rate": 0.00014888888888888886,
+      "loss": 6.3407,
+      "step": 67
+    },
+    {
+      "epoch": 0.15085967831392125,
+      "grad_norm": 1.7461535930633545,
+      "learning_rate": 0.00015111111111111109,
+      "loss": 6.2734,
+      "step": 68
+    },
+    {
+      "epoch": 0.15307820299500832,
+      "grad_norm": 2.3190548419952393,
+      "learning_rate": 0.0001533333333333333,
+      "loss": 6.1248,
+      "step": 69
+    },
+    {
+      "epoch": 0.1552967276760954,
+      "grad_norm": 1.932100772857666,
+      "learning_rate": 0.00015555555555555554,
+      "loss": 6.1844,
+      "step": 70
+    },
+    {
+      "epoch": 0.15751525235718247,
+      "grad_norm": 1.9892686605453491,
+      "learning_rate": 0.00015777777777777776,
+      "loss": 6.3337,
+      "step": 71
+    },
+    {
+      "epoch": 0.15973377703826955,
+      "grad_norm": 1.996738076210022,
+      "learning_rate": 0.00015999999999999999,
+      "loss": 6.2516,
+      "step": 72
+    },
+    {
+      "epoch": 0.16195230171935662,
+      "grad_norm": 1.7729567289352417,
+      "learning_rate": 0.0001622222222222222,
+      "loss": 6.2537,
+      "step": 73
+    },
+    {
+      "epoch": 0.1641708264004437,
+      "grad_norm": 2.0228068828582764,
+      "learning_rate": 0.00016444444444444444,
+      "loss": 6.4105,
+      "step": 74
+    },
+    {
+      "epoch": 0.16638935108153077,
+      "grad_norm": 2.1054394245147705,
+      "learning_rate": 0.00016666666666666666,
+      "loss": 6.0495,
+      "step": 75
+    },
+    {
+      "epoch": 0.16860787576261785,
+      "grad_norm": 1.7938268184661865,
+      "learning_rate": 0.00016888888888888889,
+      "loss": 6.6331,
+      "step": 76
+    },
+    {
+      "epoch": 0.17082640044370492,
+      "grad_norm": 1.801116943359375,
+      "learning_rate": 0.0001711111111111111,
+      "loss": 6.248,
+      "step": 77
+    },
+    {
+      "epoch": 0.17304492512479203,
+      "grad_norm": 2.2520036697387695,
+      "learning_rate": 0.0001733333333333333,
+      "loss": 6.1265,
+      "step": 78
+    },
+    {
+      "epoch": 0.1752634498058791,
+      "grad_norm": 2.5786898136138916,
+      "learning_rate": 0.00017555555555555553,
+      "loss": 6.4172,
+      "step": 79
+    },
+    {
+      "epoch": 0.17748197448696618,
+      "grad_norm": 2.454434394836426,
+      "learning_rate": 0.00017777777777777776,
+      "loss": 6.2877,
+      "step": 80
+    },
+    {
+      "epoch": 0.17970049916805325,
+      "grad_norm": 1.8599268198013306,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 6.2774,
+      "step": 81
+    },
+    {
+      "epoch": 0.18191902384914033,
+      "grad_norm": 3.30815052986145,
+      "learning_rate": 0.0001822222222222222,
+      "loss": 6.2997,
+      "step": 82
+    },
+    {
+      "epoch": 0.1841375485302274,
+      "grad_norm": 2.632484197616577,
+      "learning_rate": 0.00018444444444444443,
+      "loss": 6.1476,
+      "step": 83
+    },
+    {
+      "epoch": 0.18635607321131448,
+      "grad_norm": 2.256540060043335,
+      "learning_rate": 0.00018666666666666666,
+      "loss": 6.1887,
+      "step": 84
+    },
+    {
+      "epoch": 0.18857459789240155,
+      "grad_norm": 2.5506718158721924,
+      "learning_rate": 0.00018888888888888888,
+      "loss": 6.1926,
+      "step": 85
+    },
+    {
+      "epoch": 0.19079312257348863,
+      "grad_norm": 2.3097774982452393,
+      "learning_rate": 0.0001911111111111111,
+      "loss": 5.961,
+      "step": 86
+    },
+    {
+      "epoch": 0.1930116472545757,
+      "grad_norm": 3.216681480407715,
+      "learning_rate": 0.00019333333333333333,
+      "loss": 6.2499,
+      "step": 87
+    },
+    {
+      "epoch": 0.19523017193566278,
+      "grad_norm": 2.977572202682495,
+      "learning_rate": 0.00019555555555555556,
+      "loss": 6.0447,
+      "step": 88
+    },
+    {
+      "epoch": 0.19744869661674985,
+      "grad_norm": 4.106575965881348,
+      "learning_rate": 0.00019777777777777776,
+      "loss": 6.3153,
+      "step": 89
+    },
+    {
+      "epoch": 0.19966722129783693,
+      "grad_norm": 2.4921114444732666,
+      "learning_rate": 0.00019999999999999998,
+      "loss": 6.1619,
+      "step": 90
+    },
+    {
+      "epoch": 0.201885745978924,
+      "grad_norm": 2.3812241554260254,
+      "learning_rate": 0.0002022222222222222,
+      "loss": 6.1412,
+      "step": 91
+    },
+    {
+      "epoch": 0.2041042706600111,
+      "grad_norm": 3.820882797241211,
+      "learning_rate": 0.00020444444444444443,
+      "loss": 6.0223,
+      "step": 92
+    },
+    {
+      "epoch": 0.20632279534109818,
+      "grad_norm": 3.3213260173797607,
+      "learning_rate": 0.00020666666666666666,
+      "loss": 6.2297,
+      "step": 93
+    },
+    {
+      "epoch": 0.20854132002218526,
+      "grad_norm": 3.2521910667419434,
+      "learning_rate": 0.00020888888888888888,
+      "loss": 6.2223,
+      "step": 94
+    },
+    {
+      "epoch": 0.21075984470327233,
+      "grad_norm": 3.3431460857391357,
+      "learning_rate": 0.0002111111111111111,
+      "loss": 6.0618,
+      "step": 95
+    },
+    {
+      "epoch": 0.2129783693843594,
+      "grad_norm": 3.6344332695007324,
+      "learning_rate": 0.00021333333333333333,
+      "loss": 6.1421,
+      "step": 96
+    },
+    {
+      "epoch": 0.21519689406544648,
+      "grad_norm": 2.366407871246338,
+      "learning_rate": 0.00021555555555555556,
+      "loss": 5.8652,
+      "step": 97
+    },
+    {
+      "epoch": 0.21741541874653356,
+      "grad_norm": 5.667245388031006,
+      "learning_rate": 0.00021777777777777778,
+      "loss": 5.9991,
+      "step": 98
+    },
+    {
+      "epoch": 0.21963394342762063,
+      "grad_norm": 3.6710238456726074,
+      "learning_rate": 0.00021999999999999995,
+      "loss": 6.1888,
+      "step": 99
+    },
+    {
+      "epoch": 0.2218524681087077,
+      "grad_norm": 7.917342185974121,
+      "learning_rate": 0.00022222222222222218,
+      "loss": 5.8296,
+      "step": 100
+    },
+    {
+      "epoch": 0.22407099278979478,
+      "grad_norm": 7.94102144241333,
+      "learning_rate": 0.0002244444444444444,
+      "loss": 5.4338,
+      "step": 101
+    },
+    {
+      "epoch": 0.22628951747088186,
+      "grad_norm": 2.975712299346924,
+      "learning_rate": 0.00022666666666666663,
+      "loss": 5.7384,
+      "step": 102
+    },
+    {
+      "epoch": 0.22850804215196893,
+      "grad_norm": 10.478729248046875,
+      "learning_rate": 0.00022888888888888885,
+      "loss": 5.9641,
+      "step": 103
+    },
+    {
+      "epoch": 0.230726566833056,
+      "grad_norm": 11.62560749053955,
+      "learning_rate": 0.00023111111111111108,
+      "loss": 5.4342,
+      "step": 104
+    },
+    {
+      "epoch": 0.23294509151414308,
+      "grad_norm": 5.349384784698486,
+      "learning_rate": 0.0002333333333333333,
+      "loss": 5.8547,
+      "step": 105
+    },
+    {
+      "epoch": 0.23516361619523019,
+      "grad_norm": 8.264240264892578,
+      "learning_rate": 0.00023555555555555553,
+      "loss": 5.7199,
+      "step": 106
+    },
+    {
+      "epoch": 0.23738214087631726,
+      "grad_norm": 12.192222595214844,
+      "learning_rate": 0.00023777777777777775,
+      "loss": 5.6058,
+      "step": 107
+    },
+    {
+      "epoch": 0.23960066555740434,
+      "grad_norm": 4.782373428344727,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 5.7377,
+      "step": 108
+    },
+    {
+      "epoch": 0.2418191902384914,
+      "grad_norm": 9.63099193572998,
+      "learning_rate": 0.0002422222222222222,
+      "loss": 5.7224,
+      "step": 109
+    },
+    {
+      "epoch": 0.24403771491957849,
+      "grad_norm": 11.105454444885254,
+      "learning_rate": 0.00024444444444444443,
+      "loss": 5.8202,
+      "step": 110
+    },
+    {
+      "epoch": 0.24625623960066556,
+      "grad_norm": 4.963940143585205,
+      "learning_rate": 0.0002466666666666666,
+      "loss": 5.7028,
+      "step": 111
+    },
+    {
+      "epoch": 0.24847476428175264,
+      "grad_norm": 9.338000297546387,
+      "learning_rate": 0.0002488888888888889,
+      "loss": 5.4165,
+      "step": 112
+    },
+    {
+      "epoch": 0.2506932889628397,
+      "grad_norm": 9.696998596191406,
+      "learning_rate": 0.0002511111111111111,
+      "loss": 5.4409,
+      "step": 113
+    },
+    {
+      "epoch": 0.2529118136439268,
+      "grad_norm": 5.044449329376221,
+      "learning_rate": 0.00025333333333333333,
+      "loss": 5.4261,
+      "step": 114
+    },
+    {
+      "epoch": 0.25513033832501386,
+      "grad_norm": 5.547409534454346,
+      "learning_rate": 0.00025555555555555553,
+      "loss": 5.3918,
+      "step": 115
+    },
+    {
+      "epoch": 0.25734886300610094,
+      "grad_norm": 5.158348083496094,
+      "learning_rate": 0.0002577777777777778,
+      "loss": 5.4578,
+      "step": 116
+    },
+    {
+      "epoch": 0.259567387687188,
+      "grad_norm": 4.160797595977783,
+      "learning_rate": 0.00026,
+      "loss": 5.3358,
+      "step": 117
+    },
+    {
+      "epoch": 0.2617859123682751,
+      "grad_norm": 7.16011381149292,
+      "learning_rate": 0.00026222222222222223,
+      "loss": 5.3624,
+      "step": 118
+    },
+    {
+      "epoch": 0.26400443704936216,
+      "grad_norm": 5.639223575592041,
+      "learning_rate": 0.00026444444444444443,
+      "loss": 5.248,
+      "step": 119
+    },
+    {
+      "epoch": 0.26622296173044924,
+      "grad_norm": 8.2030029296875,
+      "learning_rate": 0.0002666666666666666,
+      "loss": 5.3996,
+      "step": 120
+    },
+    {
+      "epoch": 0.2684414864115363,
+      "grad_norm": 6.017608165740967,
+      "learning_rate": 0.0002688888888888889,
+      "loss": 5.0931,
+      "step": 121
+    },
+    {
+      "epoch": 0.2706600110926234,
+      "grad_norm": 9.665661811828613,
+      "learning_rate": 0.0002711111111111111,
+      "loss": 5.3154,
+      "step": 122
+    },
+    {
+      "epoch": 0.27287853577371046,
+      "grad_norm": 8.194005012512207,
+      "learning_rate": 0.00027333333333333333,
+      "loss": 5.1912,
+      "step": 123
+    },
+    {
+      "epoch": 0.27509706045479754,
+      "grad_norm": 7.089859485626221,
+      "learning_rate": 0.0002755555555555555,
+      "loss": 5.2339,
+      "step": 124
+    },
+    {
+      "epoch": 0.2773155851358846,
+      "grad_norm": 11.654187202453613,
+      "learning_rate": 0.0002777777777777778,
+      "loss": 5.4888,
+      "step": 125
+    },
+    {
+      "epoch": 0.2795341098169717,
+      "grad_norm": 9.684091567993164,
+      "learning_rate": 0.00028,
+      "loss": 5.282,
+      "step": 126
+    },
+    {
+      "epoch": 0.2817526344980588,
+      "grad_norm": 10.477103233337402,
+      "learning_rate": 0.00028222222222222223,
+      "loss": 5.205,
+      "step": 127
+    },
+    {
+      "epoch": 0.2839711591791459,
+      "grad_norm": 8.727799415588379,
+      "learning_rate": 0.0002844444444444444,
+      "loss": 5.4352,
+      "step": 128
+    },
+    {
+      "epoch": 0.28618968386023297,
+      "grad_norm": 8.56086254119873,
+      "learning_rate": 0.0002866666666666667,
+      "loss": 5.2909,
+      "step": 129
+    },
+    {
+      "epoch": 0.28840820854132004,
+      "grad_norm": 8.4285249710083,
+      "learning_rate": 0.0002888888888888888,
+      "loss": 5.3951,
+      "step": 130
+    },
+    {
+      "epoch": 0.2906267332224071,
+      "grad_norm": 6.3976054191589355,
+      "learning_rate": 0.0002911111111111111,
+      "loss": 5.4093,
+      "step": 131
+    },
+    {
+      "epoch": 0.2928452579034942,
+      "grad_norm": 6.13315486907959,
+      "learning_rate": 0.00029333333333333327,
+      "loss": 4.922,
+      "step": 132
+    },
+    {
+      "epoch": 0.29506378258458127,
+      "grad_norm": 6.330722808837891,
+      "learning_rate": 0.0002955555555555555,
+      "loss": 5.2876,
+      "step": 133
+    },
+    {
+      "epoch": 0.29728230726566834,
+      "grad_norm": 6.6806135177612305,
+      "learning_rate": 0.0002977777777777777,
+      "loss": 5.051,
+      "step": 134
+    },
+    {
+      "epoch": 0.2995008319467554,
+      "grad_norm": 5.428084373474121,
+      "learning_rate": 0.0003,
+      "loss": 4.9329,
+      "step": 135
+    },
+    {
+      "epoch": 0.3017193566278425,
+      "grad_norm": 6.041404724121094,
+      "learning_rate": 0.00029975308641975307,
+      "loss": 4.8975,
+      "step": 136
+    },
+    {
+      "epoch": 0.30393788130892957,
+      "grad_norm": 4.520870685577393,
+      "learning_rate": 0.00029950617283950616,
+      "loss": 5.3007,
+      "step": 137
+    },
+    {
+      "epoch": 0.30615640599001664,
+      "grad_norm": 9.24893856048584,
+      "learning_rate": 0.00029925925925925926,
+      "loss": 5.134,
+      "step": 138
+    },
+    {
+      "epoch": 0.3083749306711037,
+      "grad_norm": 6.243694305419922,
+      "learning_rate": 0.0002990123456790123,
+      "loss": 5.0237,
+      "step": 139
+    },
+    {
+      "epoch": 0.3105934553521908,
+      "grad_norm": 4.546097278594971,
+      "learning_rate": 0.0002987654320987654,
+      "loss": 5.0313,
+      "step": 140
+    },
+    {
+      "epoch": 0.31281198003327787,
+      "grad_norm": 5.99699592590332,
+      "learning_rate": 0.0002985185185185185,
+      "loss": 4.8896,
+      "step": 141
+    },
+    {
+      "epoch": 0.31503050471436495,
+      "grad_norm": 8.150283813476562,
+      "learning_rate": 0.0002982716049382716,
+      "loss": 5.2722,
+      "step": 142
+    },
+    {
+      "epoch": 0.317249029395452,
+      "grad_norm": 7.2810258865356445,
+      "learning_rate": 0.0002980246913580247,
+      "loss": 4.9425,
+      "step": 143
+    },
+    {
+      "epoch": 0.3194675540765391,
+      "grad_norm": 4.84526252746582,
+      "learning_rate": 0.0002977777777777777,
+      "loss": 5.1888,
+      "step": 144
+    },
+    {
+      "epoch": 0.32168607875762617,
+      "grad_norm": 6.3347954750061035,
+      "learning_rate": 0.00029753086419753087,
+      "loss": 5.0802,
+      "step": 145
+    },
+    {
+      "epoch": 0.32390460343871325,
+      "grad_norm": 5.987547397613525,
+      "learning_rate": 0.0002972839506172839,
+      "loss": 4.9879,
+      "step": 146
+    },
+    {
+      "epoch": 0.3261231281198003,
+      "grad_norm": 5.489535331726074,
+      "learning_rate": 0.000297037037037037,
+      "loss": 4.7362,
+      "step": 147
+    },
+    {
+      "epoch": 0.3283416528008874,
+      "grad_norm": 9.252480506896973,
+      "learning_rate": 0.0002967901234567901,
+      "loss": 5.2288,
+      "step": 148
+    },
+    {
+      "epoch": 0.33056017748197447,
+      "grad_norm": 8.22285270690918,
+      "learning_rate": 0.0002965432098765432,
+      "loss": 5.0849,
+      "step": 149
+    },
+    {
+      "epoch": 0.33277870216306155,
+      "grad_norm": 5.304591655731201,
+      "learning_rate": 0.00029629629629629624,
+      "loss": 4.9756,
+      "step": 150
+    },
+    {
+      "epoch": 0.3349972268441486,
+      "grad_norm": 7.345332622528076,
+      "learning_rate": 0.00029604938271604933,
+      "loss": 5.1705,
+      "step": 151
+    },
+    {
+      "epoch": 0.3372157515252357,
+      "grad_norm": 4.644253253936768,
+      "learning_rate": 0.0002958024691358025,
+      "loss": 5.0919,
+      "step": 152
+    },
+    {
+      "epoch": 0.33943427620632277,
+      "grad_norm": 3.943446636199951,
+      "learning_rate": 0.0002955555555555555,
+      "loss": 4.9817,
+      "step": 153
+    },
+    {
+      "epoch": 0.34165280088740985,
+      "grad_norm": 4.669249057769775,
+      "learning_rate": 0.0002953086419753086,
+      "loss": 4.9341,
+      "step": 154
+    },
+    {
+      "epoch": 0.343871325568497,
+      "grad_norm": 5.717892646789551,
+      "learning_rate": 0.0002950617283950617,
+      "loss": 5.1695,
+      "step": 155
+    },
+    {
+      "epoch": 0.34608985024958405,
+      "grad_norm": 4.880568027496338,
+      "learning_rate": 0.0002948148148148148,
+      "loss": 4.9563,
+      "step": 156
+    },
+    {
+      "epoch": 0.3483083749306711,
+      "grad_norm": 4.0522990226745605,
+      "learning_rate": 0.00029456790123456785,
+      "loss": 5.0349,
+      "step": 157
+    },
+    {
+      "epoch": 0.3505268996117582,
+      "grad_norm": 5.536722660064697,
+      "learning_rate": 0.00029432098765432095,
+      "loss": 5.1229,
+      "step": 158
+    },
+    {
+      "epoch": 0.3527454242928453,
+      "grad_norm": 3.4449219703674316,
+      "learning_rate": 0.00029407407407407404,
+      "loss": 4.5567,
+      "step": 159
+    },
+    {
+      "epoch": 0.35496394897393235,
+      "grad_norm": 4.434334754943848,
+      "learning_rate": 0.00029382716049382714,
+      "loss": 4.7849,
+      "step": 160
+    },
+    {
+      "epoch": 0.35718247365501943,
+      "grad_norm": 5.1075544357299805,
+      "learning_rate": 0.00029358024691358023,
+      "loss": 5.3021,
+      "step": 161
+    },
+    {
+      "epoch": 0.3594009983361065,
+      "grad_norm": 5.96935510635376,
+      "learning_rate": 0.00029333333333333327,
+      "loss": 5.0779,
+      "step": 162
+    },
+    {
+      "epoch": 0.3616195230171936,
+      "grad_norm": 6.567812919616699,
+      "learning_rate": 0.0002930864197530864,
+      "loss": 5.0314,
+      "step": 163
+    },
+    {
+      "epoch": 0.36383804769828065,
+      "grad_norm": 5.427769184112549,
+      "learning_rate": 0.00029283950617283946,
+      "loss": 4.9587,
+      "step": 164
+    },
+    {
+      "epoch": 0.36605657237936773,
+      "grad_norm": 7.360152721405029,
+      "learning_rate": 0.00029259259259259256,
+      "loss": 4.966,
+      "step": 165
+    },
+    {
+      "epoch": 0.3682750970604548,
+      "grad_norm": 5.89814567565918,
+      "learning_rate": 0.00029234567901234565,
+      "loss": 5.0995,
+      "step": 166
+    },
+    {
+      "epoch": 0.3704936217415419,
+      "grad_norm": 5.960623741149902,
+      "learning_rate": 0.00029209876543209875,
+      "loss": 4.7687,
+      "step": 167
+    },
+    {
+      "epoch": 0.37271214642262895,
+      "grad_norm": 4.919917583465576,
+      "learning_rate": 0.00029185185185185184,
+      "loss": 4.8205,
+      "step": 168
+    },
+    {
+      "epoch": 0.37493067110371603,
+      "grad_norm": 6.094188690185547,
+      "learning_rate": 0.0002916049382716049,
+      "loss": 4.667,
+      "step": 169
+    },
+    {
+      "epoch": 0.3771491957848031,
+      "grad_norm": 5.920214653015137,
+      "learning_rate": 0.000291358024691358,
+      "loss": 4.7728,
+      "step": 170
+    },
+    {
+      "epoch": 0.3793677204658902,
+      "grad_norm": 5.14661169052124,
+      "learning_rate": 0.0002911111111111111,
+      "loss": 5.0354,
+      "step": 171
+    },
+    {
+      "epoch": 0.38158624514697725,
+      "grad_norm": 5.48258638381958,
+      "learning_rate": 0.00029086419753086417,
+      "loss": 4.5832,
+      "step": 172
+    },
+    {
+      "epoch": 0.38380476982806433,
+      "grad_norm": 5.962825775146484,
+      "learning_rate": 0.00029061728395061726,
+      "loss": 4.5747,
+      "step": 173
+    },
+    {
+      "epoch": 0.3860232945091514,
+      "grad_norm": 4.948214530944824,
+      "learning_rate": 0.00029037037037037036,
+      "loss": 4.6849,
+      "step": 174
+    },
+    {
+      "epoch": 0.3882418191902385,
+      "grad_norm": 4.514078140258789,
+      "learning_rate": 0.00029012345679012345,
+      "loss": 4.8026,
+      "step": 175
+    },
+    {
+      "epoch": 0.39046034387132555,
+      "grad_norm": 4.841763496398926,
+      "learning_rate": 0.0002898765432098765,
+      "loss": 4.7999,
+      "step": 176
+    },
+    {
+      "epoch": 0.39267886855241263,
+      "grad_norm": 4.906101226806641,
+      "learning_rate": 0.0002896296296296296,
+      "loss": 4.5852,
+      "step": 177
+    },
+    {
+      "epoch": 0.3948973932334997,
+      "grad_norm": 5.288046836853027,
+      "learning_rate": 0.0002893827160493827,
+      "loss": 4.7181,
+      "step": 178
+    },
+    {
+      "epoch": 0.3971159179145868,
+      "grad_norm": 3.973954916000366,
+      "learning_rate": 0.0002891358024691358,
+      "loss": 5.0714,
+      "step": 179
+    },
+    {
+      "epoch": 0.39933444259567386,
+      "grad_norm": 3.124938726425171,
+      "learning_rate": 0.0002888888888888888,
+      "loss": 4.7596,
+      "step": 180
+    },
+    {
+      "epoch": 0.40155296727676093,
+      "grad_norm": 4.0433526039123535,
+      "learning_rate": 0.00028864197530864197,
+      "loss": 4.6988,
+      "step": 181
+    },
+    {
+      "epoch": 0.403771491957848,
+      "grad_norm": 3.744964599609375,
+      "learning_rate": 0.00028839506172839507,
+      "loss": 4.8237,
+      "step": 182
+    },
+    {
+      "epoch": 0.4059900166389351,
+      "grad_norm": 3.4403107166290283,
+      "learning_rate": 0.0002881481481481481,
+      "loss": 4.4361,
+      "step": 183
+    },
+    {
+      "epoch": 0.4082085413200222,
+      "grad_norm": 3.6703860759735107,
+      "learning_rate": 0.0002879012345679012,
+      "loss": 4.823,
+      "step": 184
+    },
+    {
+      "epoch": 0.4104270660011093,
+      "grad_norm": 3.8182108402252197,
+      "learning_rate": 0.0002876543209876543,
+      "loss": 4.7147,
+      "step": 185
+    },
+    {
+      "epoch": 0.41264559068219636,
+      "grad_norm": 4.189947128295898,
+      "learning_rate": 0.0002874074074074074,
+      "loss": 4.653,
+      "step": 186
+    },
+    {
+      "epoch": 0.41486411536328344,
+      "grad_norm": 4.308079719543457,
+      "learning_rate": 0.00028716049382716043,
+      "loss": 4.5538,
+      "step": 187
+    },
+    {
+      "epoch": 0.4170826400443705,
+      "grad_norm": 5.259140968322754,
+      "learning_rate": 0.00028691358024691353,
+      "loss": 4.7145,
+      "step": 188
+    },
+    {
+      "epoch": 0.4193011647254576,
+      "grad_norm": 3.8257129192352295,
+      "learning_rate": 0.0002866666666666667,
+      "loss": 4.6911,
+      "step": 189
+    },
+    {
+      "epoch": 0.42151968940654466,
+      "grad_norm": 3.2963294982910156,
+      "learning_rate": 0.0002864197530864197,
+      "loss": 4.5572,
+      "step": 190
+    },
+    {
+      "epoch": 0.42373821408763174,
+      "grad_norm": 4.685267448425293,
+      "learning_rate": 0.0002861728395061728,
+      "loss": 4.7226,
+      "step": 191
+    },
+    {
+      "epoch": 0.4259567387687188,
+      "grad_norm": 3.973348617553711,
+      "learning_rate": 0.0002859259259259259,
+      "loss": 4.5771,
+      "step": 192
+    },
+    {
+      "epoch": 0.4281752634498059,
+      "grad_norm": 4.025948524475098,
+      "learning_rate": 0.000285679012345679,
+      "loss": 4.7351,
+      "step": 193
+    },
+    {
+      "epoch": 0.43039378813089296,
+      "grad_norm": 4.094041347503662,
+      "learning_rate": 0.00028543209876543204,
+      "loss": 4.5677,
+      "step": 194
+    },
+    {
+      "epoch": 0.43261231281198004,
+      "grad_norm": 3.796872138977051,
+      "learning_rate": 0.00028518518518518514,
+      "loss": 4.6881,
+      "step": 195
+    },
+    {
+      "epoch": 0.4348308374930671,
+      "grad_norm": 3.574305295944214,
+      "learning_rate": 0.00028493827160493824,
+      "loss": 4.6423,
+      "step": 196
+    },
+    {
+      "epoch": 0.4370493621741542,
+      "grad_norm": 3.4384846687316895,
+      "learning_rate": 0.00028469135802469133,
+      "loss": 4.717,
+      "step": 197
+    },
+    {
+      "epoch": 0.43926788685524126,
+      "grad_norm": 4.367988586425781,
+      "learning_rate": 0.0002844444444444444,
+      "loss": 4.8152,
+      "step": 198
+    },
+    {
+      "epoch": 0.44148641153632834,
+      "grad_norm": 4.052916049957275,
+      "learning_rate": 0.0002841975308641975,
+      "loss": 4.6506,
+      "step": 199
+    },
+    {
+      "epoch": 0.4437049362174154,
+      "grad_norm": 3.9602224826812744,
+      "learning_rate": 0.0002839506172839506,
+      "loss": 4.5315,
+      "step": 200
+    },
+    {
+      "epoch": 0.4459234608985025,
+      "grad_norm": 3.615278482437134,
+      "learning_rate": 0.00028370370370370366,
+      "loss": 4.5683,
+      "step": 201
+    },
+    {
+      "epoch": 0.44814198557958956,
+      "grad_norm": 3.5161917209625244,
+      "learning_rate": 0.00028345679012345675,
+      "loss": 4.6369,
+      "step": 202
+    },
+    {
+      "epoch": 0.45036051026067664,
+      "grad_norm": 4.783361434936523,
+      "learning_rate": 0.00028320987654320985,
+      "loss": 4.8378,
+      "step": 203
+    },
+    {
+      "epoch": 0.4525790349417637,
+      "grad_norm": 4.731501579284668,
+      "learning_rate": 0.00028296296296296294,
+      "loss": 5.0393,
+      "step": 204
+    },
+    {
+      "epoch": 0.4547975596228508,
+      "grad_norm": 4.048299312591553,
+      "learning_rate": 0.00028271604938271604,
+      "loss": 4.7861,
+      "step": 205
+    },
+    {
+      "epoch": 0.45701608430393786,
+      "grad_norm": 3.896235466003418,
+      "learning_rate": 0.0002824691358024691,
+      "loss": 4.3884,
+      "step": 206
+    },
+    {
+      "epoch": 0.45923460898502494,
+      "grad_norm": 4.3430914878845215,
+      "learning_rate": 0.00028222222222222223,
+      "loss": 4.5651,
+      "step": 207
+    },
+    {
+      "epoch": 0.461453133666112,
+      "grad_norm": 3.502776861190796,
+      "learning_rate": 0.00028197530864197527,
+      "loss": 4.163,
+      "step": 208
+    },
+    {
+      "epoch": 0.4636716583471991,
+      "grad_norm": 2.9143011569976807,
+      "learning_rate": 0.00028172839506172836,
+      "loss": 4.4447,
+      "step": 209
+    },
+    {
+      "epoch": 0.46589018302828616,
+      "grad_norm": 2.876685857772827,
+      "learning_rate": 0.00028148148148148146,
+      "loss": 4.3884,
+      "step": 210
+    },
+    {
+      "epoch": 0.46810870770937324,
+      "grad_norm": 3.665203332901001,
+      "learning_rate": 0.00028123456790123455,
+      "loss": 4.6588,
+      "step": 211
+    },
+    {
+      "epoch": 0.47032723239046037,
+      "grad_norm": 3.2178902626037598,
+      "learning_rate": 0.00028098765432098765,
+      "loss": 4.1598,
+      "step": 212
+    },
+    {
+      "epoch": 0.47254575707154745,
+      "grad_norm": 3.8916146755218506,
+      "learning_rate": 0.0002807407407407407,
+      "loss": 4.6284,
+      "step": 213
+    },
+    {
+      "epoch": 0.4747642817526345,
+      "grad_norm": 2.976680040359497,
+      "learning_rate": 0.0002804938271604938,
+      "loss": 4.2686,
+      "step": 214
+    },
+    {
+      "epoch": 0.4769828064337216,
+      "grad_norm": 4.095980644226074,
+      "learning_rate": 0.0002802469135802469,
+      "loss": 4.6106,
+      "step": 215
+    },
+    {
+      "epoch": 0.47920133111480867,
+      "grad_norm": 3.285442352294922,
+      "learning_rate": 0.00028,
+      "loss": 4.3409,
+      "step": 216
+    },
+    {
+      "epoch": 0.48141985579589575,
+      "grad_norm": 3.4412479400634766,
+      "learning_rate": 0.00027975308641975307,
+      "loss": 4.4275,
+      "step": 217
+    },
+    {
+      "epoch": 0.4836383804769828,
+      "grad_norm": 3.4244863986968994,
+      "learning_rate": 0.00027950617283950617,
+      "loss": 4.2365,
+      "step": 218
+    },
+    {
+      "epoch": 0.4858569051580699,
+      "grad_norm": 3.517704963684082,
+      "learning_rate": 0.00027925925925925926,
+      "loss": 4.2484,
+      "step": 219
+    },
+    {
+      "epoch": 0.48807542983915697,
+      "grad_norm": 3.351980447769165,
+      "learning_rate": 0.0002790123456790123,
+      "loss": 4.4781,
+      "step": 220
+    },
+    {
+      "epoch": 0.49029395452024405,
+      "grad_norm": 3.1581385135650635,
+      "learning_rate": 0.0002787654320987654,
+      "loss": 4.7049,
+      "step": 221
+    },
+    {
+      "epoch": 0.4925124792013311,
+      "grad_norm": 3.665536880493164,
+      "learning_rate": 0.0002785185185185185,
+      "loss": 4.4996,
+      "step": 222
+    },
+    {
+      "epoch": 0.4947310038824182,
+      "grad_norm": 3.0067360401153564,
+      "learning_rate": 0.0002782716049382716,
+      "loss": 4.6278,
+      "step": 223
+    },
+    {
+      "epoch": 0.49694952856350527,
+      "grad_norm": 3.3070037364959717,
+      "learning_rate": 0.00027802469135802463,
+      "loss": 4.3575,
+      "step": 224
+    },
+    {
+      "epoch": 0.49916805324459235,
+      "grad_norm": 3.4161312580108643,
+      "learning_rate": 0.0002777777777777778,
+      "loss": 4.3361,
+      "step": 225
+    },
+    {
+      "epoch": 0.5013865779256794,
+      "grad_norm": 3.769873857498169,
+      "learning_rate": 0.00027753086419753087,
+      "loss": 4.5016,
+      "step": 226
+    },
+    {
+      "epoch": 0.5036051026067665,
+      "grad_norm": 3.9044601917266846,
+      "learning_rate": 0.0002772839506172839,
+      "loss": 4.6747,
+      "step": 227
+    },
+    {
+      "epoch": 0.5058236272878536,
+      "grad_norm": 3.0434083938598633,
+      "learning_rate": 0.000277037037037037,
+      "loss": 4.4198,
+      "step": 228
+    },
+    {
+      "epoch": 0.5080421519689406,
+      "grad_norm": 3.933364152908325,
+      "learning_rate": 0.0002767901234567901,
+      "loss": 4.5966,
+      "step": 229
+    },
+    {
+      "epoch": 0.5102606766500277,
+      "grad_norm": 3.918617010116577,
+      "learning_rate": 0.0002765432098765432,
+      "loss": 4.4805,
+      "step": 230
+    },
+    {
+      "epoch": 0.5124792013311148,
+      "grad_norm": 3.4121551513671875,
+      "learning_rate": 0.00027629629629629624,
+      "loss": 4.4374,
+      "step": 231
+    },
+    {
+      "epoch": 0.5146977260122019,
+      "grad_norm": 3.162271738052368,
+      "learning_rate": 0.00027604938271604934,
+      "loss": 4.2865,
+      "step": 232
+    },
+    {
+      "epoch": 0.516916250693289,
+      "grad_norm": 3.218688726425171,
+      "learning_rate": 0.00027580246913580243,
+      "loss": 4.2689,
+      "step": 233
+    },
+    {
+      "epoch": 0.519134775374376,
+      "grad_norm": 3.184856653213501,
+      "learning_rate": 0.0002755555555555555,
+      "loss": 4.2091,
+      "step": 234
+    },
+    {
+      "epoch": 0.5213533000554631,
+      "grad_norm": 2.9536099433898926,
+      "learning_rate": 0.0002753086419753086,
+      "loss": 4.2193,
+      "step": 235
+    },
+    {
+      "epoch": 0.5235718247365502,
+      "grad_norm": 2.7904129028320312,
+      "learning_rate": 0.0002750617283950617,
+      "loss": 4.4771,
+      "step": 236
+    },
+    {
+      "epoch": 0.5257903494176372,
+      "grad_norm": 3.6782636642456055,
+      "learning_rate": 0.0002748148148148148,
+      "loss": 4.4313,
+      "step": 237
+    },
+    {
+      "epoch": 0.5280088740987243,
+      "grad_norm": 2.807886838912964,
+      "learning_rate": 0.00027456790123456785,
+      "loss": 4.0572,
+      "step": 238
+    },
+    {
+      "epoch": 0.5302273987798114,
+      "grad_norm": 2.838491678237915,
+      "learning_rate": 0.00027432098765432095,
+      "loss": 4.1853,
+      "step": 239
+    },
+    {
+      "epoch": 0.5324459234608985,
+      "grad_norm": 4.175128936767578,
+      "learning_rate": 0.00027407407407407404,
+      "loss": 4.2905,
+      "step": 240
+    },
+    {
+      "epoch": 0.5346644481419855,
+      "grad_norm": 2.7108681201934814,
+      "learning_rate": 0.00027382716049382714,
+      "loss": 3.9833,
+      "step": 241
+    },
+    {
+      "epoch": 0.5368829728230726,
+      "grad_norm": 3.0116233825683594,
+      "learning_rate": 0.00027358024691358023,
+      "loss": 4.2097,
+      "step": 242
+    },
+    {
+      "epoch": 0.5391014975041597,
+      "grad_norm": 3.456681966781616,
+      "learning_rate": 0.00027333333333333333,
+      "loss": 4.2071,
+      "step": 243
+    },
+    {
+      "epoch": 0.5413200221852468,
+      "grad_norm": 3.2366092205047607,
+      "learning_rate": 0.0002730864197530864,
+      "loss": 4.317,
+      "step": 244
+    },
+    {
+      "epoch": 0.5435385468663338,
+      "grad_norm": 3.1585254669189453,
+      "learning_rate": 0.00027283950617283946,
+      "loss": 4.1328,
+      "step": 245
+    },
+    {
+      "epoch": 0.5457570715474209,
+      "grad_norm": 3.1245412826538086,
+      "learning_rate": 0.00027259259259259256,
+      "loss": 4.259,
+      "step": 246
+    },
+    {
+      "epoch": 0.547975596228508,
+      "grad_norm": 2.5358755588531494,
+      "learning_rate": 0.00027234567901234565,
+      "loss": 4.2677,
+      "step": 247
+    },
+    {
+      "epoch": 0.5501941209095951,
+      "grad_norm": 2.9378318786621094,
+      "learning_rate": 0.00027209876543209875,
+      "loss": 4.2211,
+      "step": 248
+    },
+    {
+      "epoch": 0.5524126455906821,
+      "grad_norm": 2.860238790512085,
+      "learning_rate": 0.00027185185185185184,
+      "loss": 4.1478,
+      "step": 249
+    },
+    {
+      "epoch": 0.5546311702717692,
+      "grad_norm": 2.983222246170044,
+      "learning_rate": 0.0002716049382716049,
+      "loss": 4.1733,
+      "step": 250
+    },
+    {
+      "epoch": 0.5568496949528563,
+      "grad_norm": 2.463801383972168,
+      "learning_rate": 0.00027135802469135803,
+      "loss": 4.2023,
+      "step": 251
+    },
+    {
+      "epoch": 0.5590682196339434,
+      "grad_norm": 3.3929178714752197,
+      "learning_rate": 0.0002711111111111111,
+      "loss": 4.3305,
+      "step": 252
+    },
+    {
+      "epoch": 0.5612867443150305,
+      "grad_norm": 2.578279972076416,
+      "learning_rate": 0.00027086419753086417,
+      "loss": 4.6116,
+      "step": 253
+    },
+    {
+      "epoch": 0.5635052689961176,
+      "grad_norm": 2.981677770614624,
+      "learning_rate": 0.00027061728395061727,
+      "loss": 4.3533,
+      "step": 254
+    },
+    {
+      "epoch": 0.5657237936772047,
+      "grad_norm": 2.5878260135650635,
+      "learning_rate": 0.00027037037037037036,
+      "loss": 4.2512,
+      "step": 255
+    },
+    {
+      "epoch": 0.5679423183582918,
+      "grad_norm": 2.729766845703125,
+      "learning_rate": 0.00027012345679012346,
+      "loss": 4.239,
+      "step": 256
+    },
+    {
+      "epoch": 0.5701608430393789,
+      "grad_norm": 2.9803342819213867,
+      "learning_rate": 0.0002698765432098765,
+      "loss": 4.1622,
+      "step": 257
+    },
+    {
+      "epoch": 0.5723793677204659,
+      "grad_norm": 2.766282558441162,
+      "learning_rate": 0.0002696296296296296,
+      "loss": 4.2715,
+      "step": 258
+    },
+    {
+      "epoch": 0.574597892401553,
+      "grad_norm": 2.50225830078125,
+      "learning_rate": 0.0002693827160493827,
+      "loss": 4.4122,
+      "step": 259
+    },
+    {
+      "epoch": 0.5768164170826401,
+      "grad_norm": 2.617880344390869,
+      "learning_rate": 0.0002691358024691358,
+      "loss": 4.4325,
+      "step": 260
+    },
+    {
+      "epoch": 0.5790349417637272,
+      "grad_norm": 2.8523898124694824,
+      "learning_rate": 0.0002688888888888889,
+      "loss": 4.4612,
+      "step": 261
+    },
+    {
+      "epoch": 0.5812534664448142,
+      "grad_norm": 3.020848274230957,
+      "learning_rate": 0.00026864197530864197,
+      "loss": 3.6898,
+      "step": 262
+    },
+    {
+      "epoch": 0.5834719911259013,
+      "grad_norm": 2.886863946914673,
+      "learning_rate": 0.000268395061728395,
+      "loss": 4.0612,
+      "step": 263
+    },
+    {
+      "epoch": 0.5856905158069884,
+      "grad_norm": 3.5875587463378906,
+      "learning_rate": 0.0002681481481481481,
+      "loss": 4.1658,
+      "step": 264
+    },
+    {
+      "epoch": 0.5879090404880755,
+      "grad_norm": 3.2371554374694824,
+      "learning_rate": 0.0002679012345679012,
+      "loss": 4.4436,
+      "step": 265
+    },
+    {
+      "epoch": 0.5901275651691625,
+      "grad_norm": 2.695497989654541,
+      "learning_rate": 0.0002676543209876543,
+      "loss": 4.3144,
+      "step": 266
+    },
+    {
+      "epoch": 0.5923460898502496,
+      "grad_norm": 2.7353312969207764,
+      "learning_rate": 0.0002674074074074074,
+      "loss": 4.3259,
+      "step": 267
+    },
+    {
+      "epoch": 0.5945646145313367,
+      "grad_norm": 3.3017914295196533,
+      "learning_rate": 0.00026716049382716043,
+      "loss": 4.4957,
+      "step": 268
+    },
+    {
+      "epoch": 0.5967831392124238,
+      "grad_norm": 2.5601134300231934,
+      "learning_rate": 0.0002669135802469136,
+      "loss": 4.593,
+      "step": 269
+    },
+    {
+      "epoch": 0.5990016638935108,
+      "grad_norm": 2.585540771484375,
+      "learning_rate": 0.0002666666666666666,
+      "loss": 4.2474,
+      "step": 270
+    },
+    {
+      "epoch": 0.6012201885745979,
+      "grad_norm": 3.03838849067688,
+      "learning_rate": 0.0002664197530864197,
+      "loss": 4.2352,
+      "step": 271
+    },
+    {
+      "epoch": 0.603438713255685,
+      "grad_norm": 3.13554048538208,
+      "learning_rate": 0.0002661728395061728,
+      "loss": 3.9886,
+      "step": 272
+    },
+    {
+      "epoch": 0.6056572379367721,
+      "grad_norm": 2.588900089263916,
+      "learning_rate": 0.0002659259259259259,
+      "loss": 4.2126,
+      "step": 273
+    },
+    {
+      "epoch": 0.6078757626178591,
+      "grad_norm": 3.3326327800750732,
+      "learning_rate": 0.000265679012345679,
+      "loss": 4.2611,
+      "step": 274
+    },
+    {
+      "epoch": 0.6100942872989462,
+      "grad_norm": 3.2144439220428467,
+      "learning_rate": 0.00026543209876543205,
+      "loss": 4.2556,
+      "step": 275
+    },
+    {
+      "epoch": 0.6123128119800333,
+      "grad_norm": 3.280874252319336,
+      "learning_rate": 0.00026518518518518514,
+      "loss": 4.3853,
+      "step": 276
+    },
+    {
+      "epoch": 0.6145313366611204,
+      "grad_norm": 3.2615296840667725,
+      "learning_rate": 0.00026493827160493824,
+      "loss": 4.3894,
+      "step": 277
+    },
+    {
+      "epoch": 0.6167498613422074,
+      "grad_norm": 2.9271762371063232,
+      "learning_rate": 0.00026469135802469133,
+      "loss": 4.0271,
+      "step": 278
+    },
+    {
+      "epoch": 0.6189683860232945,
+      "grad_norm": 2.4938979148864746,
+      "learning_rate": 0.00026444444444444443,
+      "loss": 4.3217,
+      "step": 279
+    },
+    {
+      "epoch": 0.6211869107043816,
+      "grad_norm": 3.7003767490386963,
+      "learning_rate": 0.0002641975308641975,
+      "loss": 4.2449,
+      "step": 280
+    },
+    {
+      "epoch": 0.6234054353854687,
+      "grad_norm": 2.4747695922851562,
+      "learning_rate": 0.0002639506172839506,
+      "loss": 4.1577,
+      "step": 281
+    },
+    {
+      "epoch": 0.6256239600665557,
+      "grad_norm": 2.3822386264801025,
+      "learning_rate": 0.00026370370370370366,
+      "loss": 3.706,
+      "step": 282
+    },
+    {
+      "epoch": 0.6278424847476428,
+      "grad_norm": 3.6372578144073486,
+      "learning_rate": 0.00026345679012345675,
+      "loss": 4.3814,
+      "step": 283
+    },
+    {
+      "epoch": 0.6300610094287299,
+      "grad_norm": 2.6979026794433594,
+      "learning_rate": 0.00026320987654320985,
+      "loss": 4.302,
+      "step": 284
+    },
+    {
+      "epoch": 0.632279534109817,
+      "grad_norm": 2.701061248779297,
+      "learning_rate": 0.00026296296296296294,
+      "loss": 4.2402,
+      "step": 285
+    },
+    {
+      "epoch": 0.634498058790904,
+      "grad_norm": 2.48140811920166,
+      "learning_rate": 0.00026271604938271604,
+      "loss": 4.1646,
+      "step": 286
+    },
+    {
+      "epoch": 0.6367165834719911,
+      "grad_norm": 2.6199264526367188,
+      "learning_rate": 0.00026246913580246913,
+      "loss": 4.2137,
+      "step": 287
+    },
+    {
+      "epoch": 0.6389351081530782,
+      "grad_norm": 2.356868028640747,
+      "learning_rate": 0.00026222222222222223,
+      "loss": 3.805,
+      "step": 288
+    },
+    {
+      "epoch": 0.6411536328341653,
+      "grad_norm": 2.2794077396392822,
+      "learning_rate": 0.00026197530864197527,
+      "loss": 4.2025,
+      "step": 289
+    },
+    {
+      "epoch": 0.6433721575152523,
+      "grad_norm": 2.9452927112579346,
+      "learning_rate": 0.00026172839506172837,
+      "loss": 3.8694,
+      "step": 290
+    },
+    {
+      "epoch": 0.6455906821963394,
+      "grad_norm": 2.6014912128448486,
+      "learning_rate": 0.00026148148148148146,
+      "loss": 4.1453,
+      "step": 291
+    },
+    {
+      "epoch": 0.6478092068774265,
+      "grad_norm": 2.2544541358947754,
+      "learning_rate": 0.00026123456790123456,
+      "loss": 4.2671,
+      "step": 292
+    },
+    {
+      "epoch": 0.6500277315585136,
+      "grad_norm": 2.3973426818847656,
+      "learning_rate": 0.0002609876543209876,
+      "loss": 4.5974,
+      "step": 293
+    },
+    {
+      "epoch": 0.6522462562396006,
+      "grad_norm": 2.8458523750305176,
+      "learning_rate": 0.0002607407407407407,
+      "loss": 4.0661,
+      "step": 294
+    },
+    {
+      "epoch": 0.6544647809206877,
+      "grad_norm": 2.7320618629455566,
+      "learning_rate": 0.00026049382716049384,
+      "loss": 4.0817,
+      "step": 295
+    },
+    {
+      "epoch": 0.6566833056017748,
+      "grad_norm": 2.60909366607666,
+      "learning_rate": 0.0002602469135802469,
+      "loss": 4.1595,
+      "step": 296
+    },
+    {
+      "epoch": 0.6589018302828619,
+      "grad_norm": 2.993919849395752,
+      "learning_rate": 0.00026,
+      "loss": 3.9928,
+      "step": 297
+    },
+    {
+      "epoch": 0.6611203549639489,
+      "grad_norm": 3.076465606689453,
+      "learning_rate": 0.00025975308641975307,
+      "loss": 4.1131,
+      "step": 298
+    },
+    {
+      "epoch": 0.663338879645036,
+      "grad_norm": 3.209273099899292,
+      "learning_rate": 0.00025950617283950617,
+      "loss": 4.3888,
+      "step": 299
+    },
+    {
+      "epoch": 0.6655574043261231,
+      "grad_norm": 2.6685333251953125,
+      "learning_rate": 0.0002592592592592592,
+      "loss": 4.2203,
+      "step": 300
+    },
+    {
+      "epoch": 0.6677759290072102,
+      "grad_norm": 2.613976240158081,
+      "learning_rate": 0.0002590123456790123,
+      "loss": 3.8263,
+      "step": 301
+    },
+    {
+      "epoch": 0.6699944536882972,
+      "grad_norm": 2.0011627674102783,
+      "learning_rate": 0.0002587654320987654,
+      "loss": 4.1556,
+      "step": 302
+    },
+    {
+      "epoch": 0.6722129783693843,
+      "grad_norm": 2.3937883377075195,
+      "learning_rate": 0.0002585185185185185,
+      "loss": 4.1367,
+      "step": 303
+    },
+    {
+      "epoch": 0.6744315030504714,
+      "grad_norm": 2.151398181915283,
+      "learning_rate": 0.0002582716049382716,
+      "loss": 4.2407,
+      "step": 304
+    },
+    {
+      "epoch": 0.6766500277315585,
+      "grad_norm": 2.6608738899230957,
+      "learning_rate": 0.0002580246913580247,
+      "loss": 4.0532,
+      "step": 305
+    },
+    {
+      "epoch": 0.6788685524126455,
+      "grad_norm": 2.8762619495391846,
+      "learning_rate": 0.0002577777777777778,
+      "loss": 4.0503,
+      "step": 306
+    },
+    {
+      "epoch": 0.6810870770937326,
+      "grad_norm": 2.173898220062256,
+      "learning_rate": 0.0002575308641975308,
+      "loss": 4.0948,
+      "step": 307
+    },
+    {
+      "epoch": 0.6833056017748197,
+      "grad_norm": 3.4465389251708984,
+      "learning_rate": 0.0002572839506172839,
+      "loss": 4.049,
+      "step": 308
+    },
+    {
+      "epoch": 0.6855241264559068,
+      "grad_norm": 2.493623733520508,
+      "learning_rate": 0.000257037037037037,
+      "loss": 3.9857,
+      "step": 309
+    },
+    {
+      "epoch": 0.687742651136994,
+      "grad_norm": 3.064328908920288,
+      "learning_rate": 0.0002567901234567901,
+      "loss": 3.8196,
+      "step": 310
+    },
+    {
+      "epoch": 0.689961175818081,
+      "grad_norm": 2.6771697998046875,
+      "learning_rate": 0.0002565432098765432,
+      "loss": 4.1799,
+      "step": 311
+    },
+    {
+      "epoch": 0.6921797004991681,
+      "grad_norm": 3.0331547260284424,
+      "learning_rate": 0.00025629629629629624,
+      "loss": 4.2406,
+      "step": 312
+    },
+    {
+      "epoch": 0.6943982251802552,
+      "grad_norm": 2.8515357971191406,
+      "learning_rate": 0.0002560493827160494,
+      "loss": 4.19,
+      "step": 313
+    },
+    {
+      "epoch": 0.6966167498613423,
+      "grad_norm": 2.3618826866149902,
+      "learning_rate": 0.00025580246913580243,
+      "loss": 3.9759,
+      "step": 314
+    },
+    {
+      "epoch": 0.6988352745424293,
+      "grad_norm": 2.2200496196746826,
+      "learning_rate": 0.00025555555555555553,
+      "loss": 4.0479,
+      "step": 315
+    },
+    {
+      "epoch": 0.7010537992235164,
+      "grad_norm": 2.7787671089172363,
+      "learning_rate": 0.0002553086419753086,
+      "loss": 3.9632,
+      "step": 316
+    },
+    {
+      "epoch": 0.7032723239046035,
+      "grad_norm": 2.386683464050293,
+      "learning_rate": 0.0002550617283950617,
+      "loss": 4.0709,
+      "step": 317
+    },
+    {
+      "epoch": 0.7054908485856906,
+      "grad_norm": 2.6233952045440674,
+      "learning_rate": 0.0002548148148148148,
+      "loss": 3.9297,
+      "step": 318
+    },
+    {
+      "epoch": 0.7077093732667776,
+      "grad_norm": 2.556239128112793,
+      "learning_rate": 0.00025456790123456785,
+      "loss": 4.0414,
+      "step": 319
+    },
+    {
+      "epoch": 0.7099278979478647,
+      "grad_norm": 2.9623236656188965,
+      "learning_rate": 0.00025432098765432095,
+      "loss": 3.9652,
+      "step": 320
+    },
+    {
+      "epoch": 0.7121464226289518,
+      "grad_norm": 2.661351442337036,
+      "learning_rate": 0.00025407407407407404,
+      "loss": 3.9982,
+      "step": 321
+    },
+    {
+      "epoch": 0.7143649473100389,
+      "grad_norm": 2.8027563095092773,
+      "learning_rate": 0.00025382716049382714,
+      "loss": 4.1641,
+      "step": 322
+    },
+    {
+      "epoch": 0.7165834719911259,
+      "grad_norm": 2.5135631561279297,
+      "learning_rate": 0.00025358024691358023,
+      "loss": 4.0505,
+      "step": 323
+    },
+    {
+      "epoch": 0.718801996672213,
+      "grad_norm": 2.3007049560546875,
+      "learning_rate": 0.00025333333333333333,
+      "loss": 4.0556,
+      "step": 324
+    },
+    {
+      "epoch": 0.7210205213533001,
+      "grad_norm": 2.8085741996765137,
+      "learning_rate": 0.0002530864197530864,
+      "loss": 3.8097,
+      "step": 325
+    },
+    {
+      "epoch": 0.7232390460343872,
+      "grad_norm": 2.8848161697387695,
+      "learning_rate": 0.00025283950617283947,
+      "loss": 4.2097,
+      "step": 326
+    },
+    {
+      "epoch": 0.7254575707154742,
+      "grad_norm": 2.466867685317993,
+      "learning_rate": 0.00025259259259259256,
+      "loss": 4.2635,
+      "step": 327
+    },
+    {
+      "epoch": 0.7276760953965613,
+      "grad_norm": 2.9695167541503906,
+      "learning_rate": 0.00025234567901234566,
+      "loss": 4.095,
+      "step": 328
+    },
+    {
+      "epoch": 0.7298946200776484,
+      "grad_norm": 2.738402843475342,
+      "learning_rate": 0.00025209876543209875,
+      "loss": 4.1931,
+      "step": 329
+    },
+    {
+      "epoch": 0.7321131447587355,
+      "grad_norm": 2.4503605365753174,
+      "learning_rate": 0.0002518518518518518,
+      "loss": 4.0214,
+      "step": 330
+    },
+    {
+      "epoch": 0.7343316694398225,
+      "grad_norm": 2.3591299057006836,
+      "learning_rate": 0.00025160493827160494,
+      "loss": 4.2562,
+      "step": 331
+    },
+    {
+      "epoch": 0.7365501941209096,
+      "grad_norm": 2.765737771987915,
+      "learning_rate": 0.00025135802469135804,
+      "loss": 4.2669,
+      "step": 332
+    },
+    {
+      "epoch": 0.7387687188019967,
+      "grad_norm": 2.3128950595855713,
+      "learning_rate": 0.0002511111111111111,
+      "loss": 3.7949,
+      "step": 333
+    },
+    {
+      "epoch": 0.7409872434830838,
+      "grad_norm": 2.3972394466400146,
+      "learning_rate": 0.00025086419753086417,
+      "loss": 4.0458,
+      "step": 334
+    },
+    {
+      "epoch": 0.7432057681641708,
+      "grad_norm": 2.530991554260254,
+      "learning_rate": 0.00025061728395061727,
+      "loss": 3.8979,
+      "step": 335
+    },
+    {
+      "epoch": 0.7454242928452579,
+      "grad_norm": 2.229652166366577,
+      "learning_rate": 0.00025037037037037036,
+      "loss": 3.8993,
+      "step": 336
+    },
+    {
+      "epoch": 0.747642817526345,
+      "grad_norm": 2.997931718826294,
+      "learning_rate": 0.0002501234567901234,
+      "loss": 4.4019,
+      "step": 337
+    },
+    {
+      "epoch": 0.7498613422074321,
+      "grad_norm": 2.2159903049468994,
+      "learning_rate": 0.0002498765432098765,
+      "loss": 4.042,
+      "step": 338
+    },
+    {
+      "epoch": 0.7520798668885191,
+      "grad_norm": 2.63863205909729,
+      "learning_rate": 0.00024962962962962965,
+      "loss": 4.2571,
+      "step": 339
+    },
+    {
+      "epoch": 0.7542983915696062,
+      "grad_norm": 2.4869332313537598,
+      "learning_rate": 0.0002493827160493827,
+      "loss": 4.2362,
+      "step": 340
+    },
+    {
+      "epoch": 0.7565169162506933,
+      "grad_norm": 2.4365644454956055,
+      "learning_rate": 0.0002491358024691358,
+      "loss": 3.8387,
+      "step": 341
+    },
+    {
+      "epoch": 0.7587354409317804,
+      "grad_norm": 2.49548077583313,
+      "learning_rate": 0.0002488888888888889,
+      "loss": 3.6874,
+      "step": 342
+    },
+    {
+      "epoch": 0.7609539656128674,
+      "grad_norm": 2.249330997467041,
+      "learning_rate": 0.000248641975308642,
+      "loss": 4.2231,
+      "step": 343
+    },
+    {
+      "epoch": 0.7631724902939545,
+      "grad_norm": 2.2270190715789795,
+      "learning_rate": 0.000248395061728395,
+      "loss": 4.1506,
+      "step": 344
+    },
+    {
+      "epoch": 0.7653910149750416,
+      "grad_norm": 2.577533483505249,
+      "learning_rate": 0.0002481481481481481,
+      "loss": 4.1181,
+      "step": 345
+    },
+    {
+      "epoch": 0.7676095396561287,
+      "grad_norm": 2.7563748359680176,
+      "learning_rate": 0.0002479012345679012,
+      "loss": 4.1223,
+      "step": 346
+    },
+    {
+      "epoch": 0.7698280643372157,
+      "grad_norm": 2.2535207271575928,
+      "learning_rate": 0.0002476543209876543,
+      "loss": 4.0788,
+      "step": 347
+    },
+    {
+      "epoch": 0.7720465890183028,
+      "grad_norm": 2.801812171936035,
+      "learning_rate": 0.0002474074074074074,
+      "loss": 4.2301,
+      "step": 348
+    },
+    {
+      "epoch": 0.7742651136993899,
+      "grad_norm": 2.3085837364196777,
+      "learning_rate": 0.0002471604938271605,
+      "loss": 3.9816,
+      "step": 349
+    },
+    {
+      "epoch": 0.776483638380477,
+      "grad_norm": 2.1942267417907715,
+      "learning_rate": 0.0002469135802469136,
+      "loss": 3.9406,
+      "step": 350
+    },
+    {
+      "epoch": 0.778702163061564,
+      "grad_norm": 2.483424663543701,
+      "learning_rate": 0.0002466666666666666,
+      "loss": 4.0134,
+      "step": 351
+    },
+    {
+      "epoch": 0.7809206877426511,
+      "grad_norm": 1.8744789361953735,
+      "learning_rate": 0.0002464197530864197,
+      "loss": 3.8927,
+      "step": 352
+    },
+    {
+      "epoch": 0.7831392124237382,
+      "grad_norm": 2.3373889923095703,
+      "learning_rate": 0.0002461728395061728,
+      "loss": 3.9949,
+      "step": 353
+    },
+    {
+      "epoch": 0.7853577371048253,
+      "grad_norm": 2.9825828075408936,
+      "learning_rate": 0.0002459259259259259,
+      "loss": 4.1332,
+      "step": 354
+    },
+    {
+      "epoch": 0.7875762617859123,
+      "grad_norm": 2.3223469257354736,
+      "learning_rate": 0.000245679012345679,
+      "loss": 4.0392,
+      "step": 355
+    },
+    {
+      "epoch": 0.7897947864669994,
+      "grad_norm": 3.2993109226226807,
+      "learning_rate": 0.00024543209876543205,
+      "loss": 4.1897,
+      "step": 356
+    },
+    {
+      "epoch": 0.7920133111480865,
+      "grad_norm": 2.5638113021850586,
+      "learning_rate": 0.0002451851851851852,
+      "loss": 3.8502,
+      "step": 357
+    },
+    {
+      "epoch": 0.7942318358291736,
+      "grad_norm": 2.5473859310150146,
+      "learning_rate": 0.00024493827160493824,
+      "loss": 3.9439,
+      "step": 358
+    },
+    {
+      "epoch": 0.7964503605102606,
+      "grad_norm": 2.2959659099578857,
+      "learning_rate": 0.00024469135802469133,
+      "loss": 3.8431,
+      "step": 359
+    },
+    {
+      "epoch": 0.7986688851913477,
+      "grad_norm": 2.6552748680114746,
+      "learning_rate": 0.00024444444444444443,
+      "loss": 4.1129,
+      "step": 360
+    },
+    {
+      "epoch": 0.8008874098724348,
+      "grad_norm": 2.206350564956665,
+      "learning_rate": 0.0002441975308641975,
+      "loss": 3.7026,
+      "step": 361
+    },
+    {
+      "epoch": 0.8031059345535219,
+      "grad_norm": 3.8386683464050293,
+      "learning_rate": 0.00024395061728395062,
+      "loss": 4.1052,
+      "step": 362
+    },
+    {
+      "epoch": 0.8053244592346089,
+      "grad_norm": 2.582000732421875,
+      "learning_rate": 0.0002437037037037037,
+      "loss": 4.2745,
+      "step": 363
+    },
+    {
+      "epoch": 0.807542983915696,
+      "grad_norm": 2.1940548419952393,
+      "learning_rate": 0.00024345679012345678,
+      "loss": 4.1447,
+      "step": 364
+    },
+    {
+      "epoch": 0.8097615085967831,
+      "grad_norm": 2.6825475692749023,
+      "learning_rate": 0.00024320987654320985,
+      "loss": 4.0255,
+      "step": 365
+    },
+    {
+      "epoch": 0.8119800332778702,
+      "grad_norm": 2.7474730014801025,
+      "learning_rate": 0.00024296296296296295,
+      "loss": 3.717,
+      "step": 366
+    },
+    {
+      "epoch": 0.8141985579589573,
+      "grad_norm": 2.607325792312622,
+      "learning_rate": 0.00024271604938271601,
+      "loss": 4.3216,
+      "step": 367
+    },
+    {
+      "epoch": 0.8164170826400444,
+      "grad_norm": 2.163196325302124,
+      "learning_rate": 0.0002424691358024691,
+      "loss": 3.8819,
+      "step": 368
+    },
+    {
+      "epoch": 0.8186356073211315,
+      "grad_norm": 2.231097459793091,
+      "learning_rate": 0.0002422222222222222,
+      "loss": 4.0228,
+      "step": 369
+    },
+    {
+      "epoch": 0.8208541320022186,
+      "grad_norm": 2.5907070636749268,
+      "learning_rate": 0.00024197530864197527,
+      "loss": 4.0289,
+      "step": 370
+    },
+    {
+      "epoch": 0.8230726566833056,
+      "grad_norm": 2.1137187480926514,
+      "learning_rate": 0.0002417283950617284,
+      "loss": 3.8022,
+      "step": 371
+    },
+    {
+      "epoch": 0.8252911813643927,
+      "grad_norm": 2.5963191986083984,
+      "learning_rate": 0.00024148148148148146,
+      "loss": 4.1475,
+      "step": 372
+    },
+    {
+      "epoch": 0.8275097060454798,
+      "grad_norm": 2.2371160984039307,
+      "learning_rate": 0.00024123456790123456,
+      "loss": 4.0048,
+      "step": 373
+    },
+    {
+      "epoch": 0.8297282307265669,
+      "grad_norm": 2.1819841861724854,
+      "learning_rate": 0.00024098765432098763,
+      "loss": 3.8568,
+      "step": 374
+    },
+    {
+      "epoch": 0.831946755407654,
+      "grad_norm": 2.511154890060425,
+      "learning_rate": 0.00024074074074074072,
+      "loss": 4.1096,
+      "step": 375
+    },
+    {
+      "epoch": 0.834165280088741,
+      "grad_norm": 2.192304849624634,
+      "learning_rate": 0.0002404938271604938,
+      "loss": 4.2213,
+      "step": 376
+    },
+    {
+      "epoch": 0.8363838047698281,
+      "grad_norm": 2.1950454711914062,
+      "learning_rate": 0.00024024691358024688,
+      "loss": 3.9923,
+      "step": 377
+    },
+    {
+      "epoch": 0.8386023294509152,
+      "grad_norm": 2.4607765674591064,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 4.1626,
+      "step": 378
+    },
+    {
+      "epoch": 0.8408208541320022,
+      "grad_norm": 2.281306266784668,
+      "learning_rate": 0.00023975308641975305,
+      "loss": 3.919,
+      "step": 379
+    },
+    {
+      "epoch": 0.8430393788130893,
+      "grad_norm": 2.5203917026519775,
+      "learning_rate": 0.00023950617283950617,
+      "loss": 3.9521,
+      "step": 380
+    },
+    {
+      "epoch": 0.8452579034941764,
+      "grad_norm": 2.204066038131714,
+      "learning_rate": 0.00023925925925925924,
+      "loss": 4.0151,
+      "step": 381
+    },
+    {
+      "epoch": 0.8474764281752635,
+      "grad_norm": 2.5320961475372314,
+      "learning_rate": 0.00023901234567901233,
+      "loss": 3.8337,
+      "step": 382
+    },
+    {
+      "epoch": 0.8496949528563505,
+      "grad_norm": 2.4367313385009766,
+      "learning_rate": 0.0002387654320987654,
+      "loss": 3.8294,
+      "step": 383
+    },
+    {
+      "epoch": 0.8519134775374376,
+      "grad_norm": 2.1221415996551514,
+      "learning_rate": 0.0002385185185185185,
+      "loss": 3.8385,
+      "step": 384
+    },
+    {
+      "epoch": 0.8541320022185247,
+      "grad_norm": 2.069728136062622,
+      "learning_rate": 0.0002382716049382716,
+      "loss": 3.9019,
+      "step": 385
+    },
+    {
+      "epoch": 0.8563505268996118,
+      "grad_norm": 3.0093424320220947,
+      "learning_rate": 0.00023802469135802466,
+      "loss": 3.9841,
+      "step": 386
+    },
+    {
+      "epoch": 0.8585690515806988,
+      "grad_norm": 2.659752368927002,
+      "learning_rate": 0.00023777777777777775,
+      "loss": 4.1823,
+      "step": 387
+    },
+    {
+      "epoch": 0.8607875762617859,
+      "grad_norm": 2.3837411403656006,
+      "learning_rate": 0.00023753086419753082,
+      "loss": 4.0403,
+      "step": 388
+    },
+    {
+      "epoch": 0.863006100942873,
+      "grad_norm": 3.2922937870025635,
+      "learning_rate": 0.00023728395061728394,
+      "loss": 4.0791,
+      "step": 389
+    },
+    {
+      "epoch": 0.8652246256239601,
+      "grad_norm": 2.5036985874176025,
+      "learning_rate": 0.000237037037037037,
+      "loss": 4.0499,
+      "step": 390
+    },
+    {
+      "epoch": 0.8674431503050472,
+      "grad_norm": 2.363783597946167,
+      "learning_rate": 0.0002367901234567901,
+      "loss": 3.9313,
+      "step": 391
+    },
+    {
+      "epoch": 0.8696616749861342,
+      "grad_norm": 2.4179019927978516,
+      "learning_rate": 0.0002365432098765432,
+      "loss": 3.8614,
+      "step": 392
+    },
+    {
+      "epoch": 0.8718801996672213,
+      "grad_norm": 2.663837432861328,
+      "learning_rate": 0.00023629629629629627,
+      "loss": 3.9839,
+      "step": 393
+    },
+    {
+      "epoch": 0.8740987243483084,
+      "grad_norm": 3.131053924560547,
+      "learning_rate": 0.00023604938271604937,
+      "loss": 4.3063,
+      "step": 394
+    },
+    {
+      "epoch": 0.8763172490293955,
+      "grad_norm": 2.0122597217559814,
+      "learning_rate": 0.00023580246913580243,
+      "loss": 3.7909,
+      "step": 395
+    },
+    {
+      "epoch": 0.8785357737104825,
+      "grad_norm": 2.3959438800811768,
+      "learning_rate": 0.00023555555555555553,
+      "loss": 4.0421,
+      "step": 396
+    },
+    {
+      "epoch": 0.8807542983915696,
+      "grad_norm": 1.9514743089675903,
+      "learning_rate": 0.0002353086419753086,
+      "loss": 4.0211,
+      "step": 397
+    },
+    {
+      "epoch": 0.8829728230726567,
+      "grad_norm": 1.9828025102615356,
+      "learning_rate": 0.00023506172839506172,
+      "loss": 3.924,
+      "step": 398
+    },
+    {
+      "epoch": 0.8851913477537438,
+      "grad_norm": 2.0926284790039062,
+      "learning_rate": 0.0002348148148148148,
+      "loss": 4.1365,
+      "step": 399
+    },
+    {
+      "epoch": 0.8874098724348308,
+      "grad_norm": 2.317344903945923,
+      "learning_rate": 0.00023456790123456788,
+      "loss": 4.126,
+      "step": 400
+    },
+    {
+      "epoch": 0.8896283971159179,
+      "grad_norm": 2.574345350265503,
+      "learning_rate": 0.00023432098765432098,
+      "loss": 3.4663,
+      "step": 401
+    },
+    {
+      "epoch": 0.891846921797005,
+      "grad_norm": 2.58981990814209,
+      "learning_rate": 0.00023407407407407405,
+      "loss": 4.0894,
+      "step": 402
+    },
+    {
+      "epoch": 0.894065446478092,
+      "grad_norm": 2.1829471588134766,
+      "learning_rate": 0.00023382716049382714,
+      "loss": 3.9552,
+      "step": 403
+    },
+    {
+      "epoch": 0.8962839711591791,
+      "grad_norm": 2.259068250656128,
+      "learning_rate": 0.0002335802469135802,
+      "loss": 3.9589,
+      "step": 404
+    },
+    {
+      "epoch": 0.8985024958402662,
+      "grad_norm": 2.110677480697632,
+      "learning_rate": 0.0002333333333333333,
+      "loss": 4.0544,
+      "step": 405
+    },
+    {
+      "epoch": 0.9007210205213533,
+      "grad_norm": 1.9347285032272339,
+      "learning_rate": 0.00023308641975308637,
+      "loss": 3.7093,
+      "step": 406
+    },
+    {
+      "epoch": 0.9029395452024404,
+      "grad_norm": 2.2097795009613037,
+      "learning_rate": 0.0002328395061728395,
+      "loss": 4.0743,
+      "step": 407
+    },
+    {
+      "epoch": 0.9051580698835274,
+      "grad_norm": 2.495041608810425,
+      "learning_rate": 0.0002325925925925926,
+      "loss": 3.9921,
+      "step": 408
+    },
+    {
+      "epoch": 0.9073765945646145,
+      "grad_norm": 2.170396327972412,
+      "learning_rate": 0.00023234567901234566,
+      "loss": 4.0774,
+      "step": 409
+    },
+    {
+      "epoch": 0.9095951192457016,
+      "grad_norm": 2.328274726867676,
+      "learning_rate": 0.00023209876543209875,
+      "loss": 3.6975,
+      "step": 410
+    },
+    {
+      "epoch": 0.9118136439267887,
+      "grad_norm": 2.552391767501831,
+      "learning_rate": 0.00023185185185185182,
+      "loss": 3.8345,
+      "step": 411
+    },
+    {
+      "epoch": 0.9140321686078757,
+      "grad_norm": 2.053554058074951,
+      "learning_rate": 0.00023160493827160492,
+      "loss": 3.7963,
+      "step": 412
+    },
+    {
+      "epoch": 0.9162506932889628,
+      "grad_norm": 2.3815441131591797,
+      "learning_rate": 0.00023135802469135798,
+      "loss": 3.7868,
+      "step": 413
+    },
+    {
+      "epoch": 0.9184692179700499,
+      "grad_norm": 3.3375606536865234,
+      "learning_rate": 0.00023111111111111108,
+      "loss": 3.9507,
+      "step": 414
+    },
+    {
+      "epoch": 0.920687742651137,
+      "grad_norm": 2.492992639541626,
+      "learning_rate": 0.0002308641975308642,
+      "loss": 3.8157,
+      "step": 415
+    },
+    {
+      "epoch": 0.922906267332224,
+      "grad_norm": 2.2680182456970215,
+      "learning_rate": 0.00023061728395061727,
+      "loss": 3.569,
+      "step": 416
+    },
+    {
+      "epoch": 0.9251247920133111,
+      "grad_norm": 3.067161798477173,
+      "learning_rate": 0.00023037037037037036,
+      "loss": 3.9325,
+      "step": 417
+    },
+    {
+      "epoch": 0.9273433166943982,
+      "grad_norm": 2.2033674716949463,
+      "learning_rate": 0.00023012345679012343,
+      "loss": 3.8774,
+      "step": 418
+    },
+    {
+      "epoch": 0.9295618413754853,
+      "grad_norm": 2.6923677921295166,
+      "learning_rate": 0.00022987654320987653,
+      "loss": 4.0329,
+      "step": 419
+    },
+    {
+      "epoch": 0.9317803660565723,
+      "grad_norm": 2.4858038425445557,
+      "learning_rate": 0.0002296296296296296,
+      "loss": 4.0175,
+      "step": 420
+    },
+    {
+      "epoch": 0.9339988907376594,
+      "grad_norm": 2.611077308654785,
+      "learning_rate": 0.0002293827160493827,
+      "loss": 3.9025,
+      "step": 421
+    },
+    {
+      "epoch": 0.9362174154187465,
+      "grad_norm": 2.186128616333008,
+      "learning_rate": 0.00022913580246913579,
+      "loss": 4.1214,
+      "step": 422
+    },
+    {
+      "epoch": 0.9384359400998337,
+      "grad_norm": 2.1582841873168945,
+      "learning_rate": 0.00022888888888888885,
+      "loss": 4.0519,
+      "step": 423
+    },
+    {
+      "epoch": 0.9406544647809207,
+      "grad_norm": 2.1829123497009277,
+      "learning_rate": 0.00022864197530864198,
+      "loss": 4.0109,
+      "step": 424
+    },
+    {
+      "epoch": 0.9428729894620078,
+      "grad_norm": 2.509711503982544,
+      "learning_rate": 0.00022839506172839504,
+      "loss": 3.9629,
+      "step": 425
+    },
+    {
+      "epoch": 0.9450915141430949,
+      "grad_norm": 2.8125388622283936,
+      "learning_rate": 0.00022814814814814814,
+      "loss": 4.1289,
+      "step": 426
+    },
+    {
+      "epoch": 0.947310038824182,
+      "grad_norm": 1.977970838546753,
+      "learning_rate": 0.0002279012345679012,
+      "loss": 3.9404,
+      "step": 427
+    },
+    {
+      "epoch": 0.949528563505269,
+      "grad_norm": 2.6768743991851807,
+      "learning_rate": 0.0002276543209876543,
+      "loss": 3.9021,
+      "step": 428
+    },
+    {
+      "epoch": 0.9517470881863561,
+      "grad_norm": 2.4110512733459473,
+      "learning_rate": 0.00022740740740740737,
+      "loss": 3.8847,
+      "step": 429
+    },
+    {
+      "epoch": 0.9539656128674432,
+      "grad_norm": 2.885075807571411,
+      "learning_rate": 0.00022716049382716047,
+      "loss": 3.7745,
+      "step": 430
+    },
+    {
+      "epoch": 0.9561841375485303,
+      "grad_norm": 2.210596799850464,
+      "learning_rate": 0.00022691358024691356,
+      "loss": 3.988,
+      "step": 431
+    },
+    {
+      "epoch": 0.9584026622296173,
+      "grad_norm": 2.4937257766723633,
+      "learning_rate": 0.00022666666666666663,
+      "loss": 3.8832,
+      "step": 432
+    },
+    {
+      "epoch": 0.9606211869107044,
+      "grad_norm": 2.5230414867401123,
+      "learning_rate": 0.00022641975308641975,
+      "loss": 3.8653,
+      "step": 433
+    },
+    {
+      "epoch": 0.9628397115917915,
+      "grad_norm": 2.4804039001464844,
+      "learning_rate": 0.00022617283950617282,
+      "loss": 4.0924,
+      "step": 434
+    },
+    {
+      "epoch": 0.9650582362728786,
+      "grad_norm": 2.55033016204834,
+      "learning_rate": 0.00022592592592592591,
+      "loss": 3.5937,
+      "step": 435
+    },
+    {
+      "epoch": 0.9672767609539656,
+      "grad_norm": 2.030961275100708,
+      "learning_rate": 0.00022567901234567898,
+      "loss": 4.0473,
+      "step": 436
+    },
+    {
+      "epoch": 0.9694952856350527,
+      "grad_norm": 2.287344217300415,
+      "learning_rate": 0.00022543209876543208,
+      "loss": 3.903,
+      "step": 437
+    },
+    {
+      "epoch": 0.9717138103161398,
+      "grad_norm": 2.3059279918670654,
+      "learning_rate": 0.00022518518518518517,
+      "loss": 3.7358,
+      "step": 438
+    },
+    {
+      "epoch": 0.9739323349972269,
+      "grad_norm": 2.2446322441101074,
+      "learning_rate": 0.00022493827160493824,
+      "loss": 3.9142,
+      "step": 439
+    },
+    {
+      "epoch": 0.9761508596783139,
+      "grad_norm": 2.2679951190948486,
+      "learning_rate": 0.00022469135802469134,
+      "loss": 3.9321,
+      "step": 440
+    },
+    {
+      "epoch": 0.978369384359401,
+      "grad_norm": 1.9643009901046753,
+      "learning_rate": 0.0002244444444444444,
+      "loss": 3.6496,
+      "step": 441
+    },
+    {
+      "epoch": 0.9805879090404881,
+      "grad_norm": 2.235593795776367,
+      "learning_rate": 0.00022419753086419753,
+      "loss": 3.6349,
+      "step": 442
+    },
+    {
+      "epoch": 0.9828064337215752,
+      "grad_norm": 2.395341396331787,
+      "learning_rate": 0.0002239506172839506,
+      "loss": 4.1796,
+      "step": 443
+    },
+    {
+      "epoch": 0.9850249584026622,
+      "grad_norm": 2.8146896362304688,
+      "learning_rate": 0.0002237037037037037,
+      "loss": 3.9112,
+      "step": 444
+    },
+    {
+      "epoch": 0.9872434830837493,
+      "grad_norm": 2.725193977355957,
+      "learning_rate": 0.00022345679012345678,
+      "loss": 3.8317,
+      "step": 445
+    },
+    {
+      "epoch": 0.9894620077648364,
+      "grad_norm": 2.8570854663848877,
+      "learning_rate": 0.00022320987654320985,
+      "loss": 3.9377,
+      "step": 446
+    },
+    {
+      "epoch": 0.9916805324459235,
+      "grad_norm": 2.241975784301758,
+      "learning_rate": 0.00022296296296296295,
+      "loss": 3.6163,
+      "step": 447
+    },
+    {
+      "epoch": 0.9938990571270105,
+      "grad_norm": 2.548020124435425,
+      "learning_rate": 0.00022271604938271602,
+      "loss": 4.0965,
+      "step": 448
+    },
+    {
+      "epoch": 0.9961175818080976,
+      "grad_norm": 2.3386425971984863,
+      "learning_rate": 0.0002224691358024691,
+      "loss": 3.6864,
+      "step": 449
+    },
+    {
+      "epoch": 0.9983361064891847,
+      "grad_norm": 2.946790933609009,
+      "learning_rate": 0.00022222222222222218,
+      "loss": 3.6545,
+      "step": 450
+    },
+    {
+      "epoch": 1.0005546311702718,
+      "grad_norm": 2.1557211875915527,
+      "learning_rate": 0.0002219753086419753,
+      "loss": 4.1233,
+      "step": 451
+    },
+    {
+      "epoch": 1.0027731558513588,
+      "grad_norm": 2.443913459777832,
+      "learning_rate": 0.0002217283950617284,
+      "loss": 3.8345,
+      "step": 452
+    },
+    {
+      "epoch": 1.004991680532446,
+      "grad_norm": 2.6182546615600586,
+      "learning_rate": 0.00022148148148148146,
+      "loss": 3.9119,
+      "step": 453
+    },
+    {
+      "epoch": 1.007210205213533,
+      "grad_norm": 2.8780674934387207,
+      "learning_rate": 0.00022123456790123456,
+      "loss": 3.8838,
+      "step": 454
+    },
+    {
+      "epoch": 1.00942872989462,
+      "grad_norm": 2.455169200897217,
+      "learning_rate": 0.00022098765432098763,
+      "loss": 3.8642,
+      "step": 455
+    },
+    {
+      "epoch": 1.0116472545757071,
+      "grad_norm": 2.7115437984466553,
+      "learning_rate": 0.00022074074074074072,
+      "loss": 3.7702,
+      "step": 456
+    },
+    {
+      "epoch": 1.0138657792567942,
+      "grad_norm": 2.527395725250244,
+      "learning_rate": 0.0002204938271604938,
+      "loss": 3.5374,
+      "step": 457
+    },
+    {
+      "epoch": 1.0160843039378813,
+      "grad_norm": 3.1381354331970215,
+      "learning_rate": 0.00022024691358024689,
+      "loss": 3.5813,
+      "step": 458
+    },
+    {
+      "epoch": 1.0183028286189684,
+      "grad_norm": 2.7199149131774902,
+      "learning_rate": 0.00021999999999999995,
+      "loss": 3.8312,
+      "step": 459
+    },
+    {
+      "epoch": 1.0205213533000554,
+      "grad_norm": 2.238412380218506,
+      "learning_rate": 0.00021975308641975308,
+      "loss": 3.5278,
+      "step": 460
+    },
+    {
+      "epoch": 1.0227398779811425,
+      "grad_norm": 2.5880277156829834,
+      "learning_rate": 0.00021950617283950617,
+      "loss": 3.7138,
+      "step": 461
+    },
+    {
+      "epoch": 1.0249584026622296,
+      "grad_norm": 2.262570858001709,
+      "learning_rate": 0.00021925925925925924,
+      "loss": 3.9182,
+      "step": 462
+    },
+    {
+      "epoch": 1.0271769273433167,
+      "grad_norm": 2.257063150405884,
+      "learning_rate": 0.00021901234567901233,
+      "loss": 3.8059,
+      "step": 463
+    },
+    {
+      "epoch": 1.0293954520244037,
+      "grad_norm": 2.023219347000122,
+      "learning_rate": 0.0002187654320987654,
+      "loss": 3.6194,
+      "step": 464
+    },
+    {
+      "epoch": 1.0316139767054908,
+      "grad_norm": 2.7524654865264893,
+      "learning_rate": 0.0002185185185185185,
+      "loss": 4.0726,
+      "step": 465
+    },
+    {
+      "epoch": 1.033832501386578,
+      "grad_norm": 2.3610594272613525,
+      "learning_rate": 0.00021827160493827157,
+      "loss": 3.7312,
+      "step": 466
+    },
+    {
+      "epoch": 1.036051026067665,
+      "grad_norm": 2.1470367908477783,
+      "learning_rate": 0.00021802469135802466,
+      "loss": 3.8591,
+      "step": 467
+    },
+    {
+      "epoch": 1.038269550748752,
+      "grad_norm": 2.5584826469421387,
+      "learning_rate": 0.00021777777777777778,
+      "loss": 3.6278,
+      "step": 468
+    },
+    {
+      "epoch": 1.0404880754298391,
+      "grad_norm": 2.644559144973755,
+      "learning_rate": 0.00021753086419753085,
+      "loss": 3.7073,
+      "step": 469
+    },
+    {
+      "epoch": 1.0427066001109262,
+      "grad_norm": 2.4035820960998535,
+      "learning_rate": 0.00021728395061728395,
+      "loss": 3.6851,
+      "step": 470
+    },
+    {
+      "epoch": 1.0449251247920133,
+      "grad_norm": 2.3119945526123047,
+      "learning_rate": 0.00021703703703703701,
+      "loss": 3.4736,
+      "step": 471
+    },
+    {
+      "epoch": 1.0471436494731003,
+      "grad_norm": 2.420701026916504,
+      "learning_rate": 0.0002167901234567901,
+      "loss": 3.8762,
+      "step": 472
+    },
+    {
+      "epoch": 1.0493621741541874,
+      "grad_norm": 2.3250856399536133,
+      "learning_rate": 0.00021654320987654318,
+      "loss": 3.6779,
+      "step": 473
+    },
+    {
+      "epoch": 1.0515806988352745,
+      "grad_norm": 2.1539576053619385,
+      "learning_rate": 0.00021629629629629627,
+      "loss": 3.9242,
+      "step": 474
+    },
+    {
+      "epoch": 1.0537992235163616,
+      "grad_norm": 2.791045665740967,
+      "learning_rate": 0.00021604938271604937,
+      "loss": 3.8468,
+      "step": 475
+    },
+    {
+      "epoch": 1.0560177481974486,
+      "grad_norm": 2.609091281890869,
+      "learning_rate": 0.00021580246913580244,
+      "loss": 3.5457,
+      "step": 476
+    },
+    {
+      "epoch": 1.0582362728785357,
+      "grad_norm": 2.566927671432495,
+      "learning_rate": 0.00021555555555555556,
+      "loss": 3.437,
+      "step": 477
+    },
+    {
+      "epoch": 1.0604547975596228,
+      "grad_norm": 2.8259754180908203,
+      "learning_rate": 0.00021530864197530863,
+      "loss": 3.4957,
+      "step": 478
+    },
+    {
+      "epoch": 1.0626733222407099,
+      "grad_norm": 2.4702820777893066,
+      "learning_rate": 0.00021506172839506172,
+      "loss": 3.9053,
+      "step": 479
+    },
+    {
+      "epoch": 1.064891846921797,
+      "grad_norm": 2.222761869430542,
+      "learning_rate": 0.0002148148148148148,
+      "loss": 3.9213,
+      "step": 480
+    },
+    {
+      "epoch": 1.067110371602884,
+      "grad_norm": 2.0323846340179443,
+      "learning_rate": 0.00021456790123456788,
+      "loss": 3.5414,
+      "step": 481
+    },
+    {
+      "epoch": 1.069328896283971,
+      "grad_norm": 2.911721706390381,
+      "learning_rate": 0.00021432098765432095,
+      "loss": 3.6684,
+      "step": 482
+    },
+    {
+      "epoch": 1.0715474209650582,
+      "grad_norm": 2.8786699771881104,
+      "learning_rate": 0.00021407407407407405,
+      "loss": 4.1322,
+      "step": 483
+    },
+    {
+      "epoch": 1.0737659456461452,
+      "grad_norm": 2.8469934463500977,
+      "learning_rate": 0.00021382716049382714,
+      "loss": 3.8723,
+      "step": 484
+    },
+    {
+      "epoch": 1.0759844703272323,
+      "grad_norm": 2.982919931411743,
+      "learning_rate": 0.0002135802469135802,
+      "loss": 3.5846,
+      "step": 485
+    },
+    {
+      "epoch": 1.0782029950083194,
+      "grad_norm": 2.9622621536254883,
+      "learning_rate": 0.00021333333333333333,
+      "loss": 4.0818,
+      "step": 486
+    },
+    {
+      "epoch": 1.0804215196894065,
+      "grad_norm": 2.767716646194458,
+      "learning_rate": 0.0002130864197530864,
+      "loss": 3.6795,
+      "step": 487
+    },
+    {
+      "epoch": 1.0826400443704935,
+      "grad_norm": 2.4765114784240723,
+      "learning_rate": 0.0002128395061728395,
+      "loss": 3.848,
+      "step": 488
+    },
+    {
+      "epoch": 1.0848585690515806,
+      "grad_norm": 2.430499315261841,
+      "learning_rate": 0.00021259259259259256,
+      "loss": 4.095,
+      "step": 489
+    },
+    {
+      "epoch": 1.0870770937326677,
+      "grad_norm": 2.3223416805267334,
+      "learning_rate": 0.00021234567901234566,
+      "loss": 3.8788,
+      "step": 490
+    },
+    {
+      "epoch": 1.0892956184137548,
+      "grad_norm": 2.6114752292633057,
+      "learning_rate": 0.00021209876543209875,
+      "loss": 3.664,
+      "step": 491
+    },
+    {
+      "epoch": 1.0915141430948418,
+      "grad_norm": 2.3392255306243896,
+      "learning_rate": 0.00021185185185185182,
+      "loss": 4.0941,
+      "step": 492
+    },
+    {
+      "epoch": 1.093732667775929,
+      "grad_norm": 2.1776010990142822,
+      "learning_rate": 0.00021160493827160492,
+      "loss": 3.8825,
+      "step": 493
+    },
+    {
+      "epoch": 1.095951192457016,
+      "grad_norm": 2.108083486557007,
+      "learning_rate": 0.00021135802469135799,
+      "loss": 3.6068,
+      "step": 494
+    },
+    {
+      "epoch": 1.098169717138103,
+      "grad_norm": 2.780726909637451,
+      "learning_rate": 0.0002111111111111111,
+      "loss": 3.7723,
+      "step": 495
+    },
+    {
+      "epoch": 1.1003882418191901,
+      "grad_norm": 2.3707642555236816,
+      "learning_rate": 0.00021086419753086418,
+      "loss": 3.7925,
+      "step": 496
+    },
+    {
+      "epoch": 1.1026067665002772,
+      "grad_norm": 2.1636648178100586,
+      "learning_rate": 0.00021061728395061727,
+      "loss": 3.9388,
+      "step": 497
+    },
+    {
+      "epoch": 1.1048252911813643,
+      "grad_norm": 2.057589054107666,
+      "learning_rate": 0.00021037037037037037,
+      "loss": 3.6733,
+      "step": 498
+    },
+    {
+      "epoch": 1.1070438158624514,
+      "grad_norm": 3.065032958984375,
+      "learning_rate": 0.00021012345679012343,
+      "loss": 3.8427,
+      "step": 499
+    },
+    {
+      "epoch": 1.1092623405435384,
+      "grad_norm": 2.5414822101593018,
+      "learning_rate": 0.00020987654320987653,
+      "loss": 3.6839,
+      "step": 500
+    },
+    {
+      "epoch": 1.1114808652246255,
+      "grad_norm": 2.988729476928711,
+      "learning_rate": 0.0002096296296296296,
+      "loss": 3.5326,
+      "step": 501
+    },
+    {
+      "epoch": 1.1136993899057126,
+      "grad_norm": 3.06571888923645,
+      "learning_rate": 0.0002093827160493827,
+      "loss": 3.8799,
+      "step": 502
+    },
+    {
+      "epoch": 1.1159179145867997,
+      "grad_norm": 2.8324365615844727,
+      "learning_rate": 0.00020913580246913576,
+      "loss": 3.9904,
+      "step": 503
+    },
+    {
+      "epoch": 1.1181364392678868,
+      "grad_norm": 3.080937147140503,
+      "learning_rate": 0.00020888888888888888,
+      "loss": 3.8155,
+      "step": 504
+    },
+    {
+      "epoch": 1.1203549639489738,
+      "grad_norm": 2.3534884452819824,
+      "learning_rate": 0.00020864197530864198,
+      "loss": 3.7706,
+      "step": 505
+    },
+    {
+      "epoch": 1.122573488630061,
+      "grad_norm": 2.3190488815307617,
+      "learning_rate": 0.00020839506172839505,
+      "loss": 3.5714,
+      "step": 506
+    },
+    {
+      "epoch": 1.124792013311148,
+      "grad_norm": 2.610675811767578,
+      "learning_rate": 0.00020814814814814814,
+      "loss": 4.0656,
+      "step": 507
+    },
+    {
+      "epoch": 1.127010537992235,
+      "grad_norm": 2.2441468238830566,
+      "learning_rate": 0.0002079012345679012,
+      "loss": 3.3638,
+      "step": 508
+    },
+    {
+      "epoch": 1.1292290626733221,
+      "grad_norm": 2.429640293121338,
+      "learning_rate": 0.0002076543209876543,
+      "loss": 3.6889,
+      "step": 509
+    },
+    {
+      "epoch": 1.1314475873544092,
+      "grad_norm": 2.3146069049835205,
+      "learning_rate": 0.00020740740740740737,
+      "loss": 3.776,
+      "step": 510
+    },
+    {
+      "epoch": 1.1336661120354963,
+      "grad_norm": 2.1505658626556396,
+      "learning_rate": 0.00020716049382716047,
+      "loss": 4.0823,
+      "step": 511
+    },
+    {
+      "epoch": 1.1358846367165834,
+      "grad_norm": 2.3604695796966553,
+      "learning_rate": 0.00020691358024691354,
+      "loss": 3.8107,
+      "step": 512
+    },
+    {
+      "epoch": 1.1381031613976704,
+      "grad_norm": 2.1294548511505127,
+      "learning_rate": 0.00020666666666666666,
+      "loss": 3.7644,
+      "step": 513
+    },
+    {
+      "epoch": 1.1403216860787575,
+      "grad_norm": 2.5928773880004883,
+      "learning_rate": 0.00020641975308641975,
+      "loss": 3.3744,
+      "step": 514
+    },
+    {
+      "epoch": 1.1425402107598448,
+      "grad_norm": 2.39784836769104,
+      "learning_rate": 0.00020617283950617282,
+      "loss": 3.6539,
+      "step": 515
+    },
+    {
+      "epoch": 1.1447587354409319,
+      "grad_norm": 2.1956284046173096,
+      "learning_rate": 0.00020592592592592592,
+      "loss": 4.0354,
+      "step": 516
+    },
+    {
+      "epoch": 1.146977260122019,
+      "grad_norm": 2.6869590282440186,
+      "learning_rate": 0.00020567901234567898,
+      "loss": 4.0057,
+      "step": 517
+    },
+    {
+      "epoch": 1.149195784803106,
+      "grad_norm": 2.9978363513946533,
+      "learning_rate": 0.00020543209876543208,
+      "loss": 3.8195,
+      "step": 518
+    },
+    {
+      "epoch": 1.151414309484193,
+      "grad_norm": 2.641946792602539,
+      "learning_rate": 0.00020518518518518515,
+      "loss": 3.912,
+      "step": 519
+    },
+    {
+      "epoch": 1.1536328341652802,
+      "grad_norm": 2.9991438388824463,
+      "learning_rate": 0.00020493827160493824,
+      "loss": 3.4105,
+      "step": 520
+    },
+    {
+      "epoch": 1.1558513588463672,
+      "grad_norm": 2.3648734092712402,
+      "learning_rate": 0.00020469135802469136,
+      "loss": 3.746,
+      "step": 521
+    },
+    {
+      "epoch": 1.1580698835274543,
+      "grad_norm": 2.3753104209899902,
+      "learning_rate": 0.00020444444444444443,
+      "loss": 3.794,
+      "step": 522
+    },
+    {
+      "epoch": 1.1602884082085414,
+      "grad_norm": 3.389559268951416,
+      "learning_rate": 0.00020419753086419753,
+      "loss": 3.8021,
+      "step": 523
+    },
+    {
+      "epoch": 1.1625069328896285,
+      "grad_norm": 2.4276576042175293,
+      "learning_rate": 0.0002039506172839506,
+      "loss": 3.8302,
+      "step": 524
+    },
+    {
+      "epoch": 1.1647254575707155,
+      "grad_norm": 2.8832006454467773,
+      "learning_rate": 0.0002037037037037037,
+      "loss": 3.6804,
+      "step": 525
+    },
+    {
+      "epoch": 1.1669439822518026,
+      "grad_norm": 2.583534002304077,
+      "learning_rate": 0.00020345679012345676,
+      "loss": 3.69,
+      "step": 526
+    },
+    {
+      "epoch": 1.1691625069328897,
+      "grad_norm": 2.239502191543579,
+      "learning_rate": 0.00020320987654320985,
+      "loss": 3.4872,
+      "step": 527
+    },
+    {
+      "epoch": 1.1713810316139768,
+      "grad_norm": 3.0505623817443848,
+      "learning_rate": 0.00020296296296296295,
+      "loss": 3.8779,
+      "step": 528
+    },
+    {
+      "epoch": 1.1735995562950639,
+      "grad_norm": 2.8697545528411865,
+      "learning_rate": 0.00020271604938271602,
+      "loss": 3.6809,
+      "step": 529
+    },
+    {
+      "epoch": 1.175818080976151,
+      "grad_norm": 2.393893003463745,
+      "learning_rate": 0.00020246913580246914,
+      "loss": 3.6401,
+      "step": 530
+    },
+    {
+      "epoch": 1.178036605657238,
+      "grad_norm": 2.81330943107605,
+      "learning_rate": 0.0002022222222222222,
+      "loss": 3.864,
+      "step": 531
+    },
+    {
+      "epoch": 1.180255130338325,
+      "grad_norm": 2.9494330883026123,
+      "learning_rate": 0.0002019753086419753,
+      "loss": 3.9188,
+      "step": 532
+    },
+    {
+      "epoch": 1.1824736550194122,
+      "grad_norm": 2.7004880905151367,
+      "learning_rate": 0.00020172839506172837,
+      "loss": 3.4328,
+      "step": 533
+    },
+    {
+      "epoch": 1.1846921797004992,
+      "grad_norm": 2.1835241317749023,
+      "learning_rate": 0.00020148148148148147,
+      "loss": 3.4772,
+      "step": 534
+    },
+    {
+      "epoch": 1.1869107043815863,
+      "grad_norm": 2.5207808017730713,
+      "learning_rate": 0.00020123456790123456,
+      "loss": 3.8006,
+      "step": 535
+    },
+    {
+      "epoch": 1.1891292290626734,
+      "grad_norm": 2.858858108520508,
+      "learning_rate": 0.00020098765432098763,
+      "loss": 3.5053,
+      "step": 536
+    },
+    {
+      "epoch": 1.1913477537437605,
+      "grad_norm": 2.916020154953003,
+      "learning_rate": 0.00020074074074074072,
+      "loss": 3.6071,
+      "step": 537
+    },
+    {
+      "epoch": 1.1935662784248475,
+      "grad_norm": 2.3896844387054443,
+      "learning_rate": 0.0002004938271604938,
+      "loss": 3.7092,
+      "step": 538
+    },
+    {
+      "epoch": 1.1957848031059346,
+      "grad_norm": 2.4803242683410645,
+      "learning_rate": 0.00020024691358024691,
+      "loss": 3.8119,
+      "step": 539
+    },
+    {
+      "epoch": 1.1980033277870217,
+      "grad_norm": 3.172659158706665,
+      "learning_rate": 0.00019999999999999998,
+      "loss": 3.7684,
+      "step": 540
+    },
+    {
+      "epoch": 1.2002218524681088,
+      "grad_norm": 2.3113019466400146,
+      "learning_rate": 0.00019975308641975308,
+      "loss": 3.5668,
+      "step": 541
+    },
+    {
+      "epoch": 1.2024403771491958,
+      "grad_norm": 2.8322885036468506,
+      "learning_rate": 0.00019950617283950615,
+      "loss": 3.6609,
+      "step": 542
+    },
+    {
+      "epoch": 1.204658901830283,
+      "grad_norm": 2.3816561698913574,
+      "learning_rate": 0.00019925925925925924,
+      "loss": 3.7563,
+      "step": 543
+    },
+    {
+      "epoch": 1.20687742651137,
+      "grad_norm": 2.3493103981018066,
+      "learning_rate": 0.00019901234567901234,
+      "loss": 3.5583,
+      "step": 544
+    },
+    {
+      "epoch": 1.209095951192457,
+      "grad_norm": 2.423032283782959,
+      "learning_rate": 0.0001987654320987654,
+      "loss": 3.6248,
+      "step": 545
+    },
+    {
+      "epoch": 1.2113144758735441,
+      "grad_norm": 3.779360055923462,
+      "learning_rate": 0.0001985185185185185,
+      "loss": 3.5377,
+      "step": 546
+    },
+    {
+      "epoch": 1.2135330005546312,
+      "grad_norm": 3.2507946491241455,
+      "learning_rate": 0.00019827160493827157,
+      "loss": 3.4752,
+      "step": 547
+    },
+    {
+      "epoch": 1.2157515252357183,
+      "grad_norm": 3.0638017654418945,
+      "learning_rate": 0.0001980246913580247,
+      "loss": 3.5649,
+      "step": 548
+    },
+    {
+      "epoch": 1.2179700499168054,
+      "grad_norm": 2.8732802867889404,
+      "learning_rate": 0.00019777777777777776,
+      "loss": 3.79,
+      "step": 549
+    },
+    {
+      "epoch": 1.2201885745978924,
+      "grad_norm": 3.06744384765625,
+      "learning_rate": 0.00019753086419753085,
+      "loss": 3.6923,
+      "step": 550
+    },
+    {
+      "epoch": 1.2224070992789795,
+      "grad_norm": 3.0948922634124756,
+      "learning_rate": 0.00019728395061728395,
+      "loss": 3.674,
+      "step": 551
+    },
+    {
+      "epoch": 1.2246256239600666,
+      "grad_norm": 3.763289451599121,
+      "learning_rate": 0.00019703703703703702,
+      "loss": 3.801,
+      "step": 552
+    },
+    {
+      "epoch": 1.2268441486411537,
+      "grad_norm": 2.7338316440582275,
+      "learning_rate": 0.0001967901234567901,
+      "loss": 3.6216,
+      "step": 553
+    },
+    {
+      "epoch": 1.2290626733222407,
+      "grad_norm": 2.985049247741699,
+      "learning_rate": 0.00019654320987654318,
+      "loss": 3.7071,
+      "step": 554
+    },
+    {
+      "epoch": 1.2312811980033278,
+      "grad_norm": 2.4293172359466553,
+      "learning_rate": 0.00019629629629629627,
+      "loss": 3.5639,
+      "step": 555
+    },
+    {
+      "epoch": 1.2334997226844149,
+      "grad_norm": 2.2493081092834473,
+      "learning_rate": 0.00019604938271604934,
+      "loss": 3.6523,
+      "step": 556
+    },
+    {
+      "epoch": 1.235718247365502,
+      "grad_norm": 2.8648602962493896,
+      "learning_rate": 0.00019580246913580246,
+      "loss": 3.6334,
+      "step": 557
+    },
+    {
+      "epoch": 1.237936772046589,
+      "grad_norm": 2.7224338054656982,
+      "learning_rate": 0.00019555555555555556,
+      "loss": 3.373,
+      "step": 558
+    },
+    {
+      "epoch": 1.240155296727676,
+      "grad_norm": 2.9226064682006836,
+      "learning_rate": 0.00019530864197530863,
+      "loss": 3.9188,
+      "step": 559
+    },
+    {
+      "epoch": 1.2423738214087632,
+      "grad_norm": 2.7861526012420654,
+      "learning_rate": 0.00019506172839506172,
+      "loss": 3.7667,
+      "step": 560
+    },
+    {
+      "epoch": 1.2445923460898503,
+      "grad_norm": 2.4156463146209717,
+      "learning_rate": 0.0001948148148148148,
+      "loss": 3.4996,
+      "step": 561
+    },
+    {
+      "epoch": 1.2468108707709373,
+      "grad_norm": 2.3606348037719727,
+      "learning_rate": 0.00019456790123456789,
+      "loss": 3.6083,
+      "step": 562
+    },
+    {
+      "epoch": 1.2490293954520244,
+      "grad_norm": 2.445591688156128,
+      "learning_rate": 0.00019432098765432095,
+      "loss": 3.4759,
+      "step": 563
+    },
+    {
+      "epoch": 1.2512479201331115,
+      "grad_norm": 2.752551317214966,
+      "learning_rate": 0.00019407407407407405,
+      "loss": 3.6908,
+      "step": 564
+    },
+    {
+      "epoch": 1.2534664448141986,
+      "grad_norm": 2.6307296752929688,
+      "learning_rate": 0.00019382716049382712,
+      "loss": 3.6304,
+      "step": 565
+    },
+    {
+      "epoch": 1.2556849694952856,
+      "grad_norm": 2.8586552143096924,
+      "learning_rate": 0.00019358024691358024,
+      "loss": 3.6859,
+      "step": 566
+    },
+    {
+      "epoch": 1.2579034941763727,
+      "grad_norm": 2.68198299407959,
+      "learning_rate": 0.00019333333333333333,
+      "loss": 3.8644,
+      "step": 567
+    },
+    {
+      "epoch": 1.2601220188574598,
+      "grad_norm": 2.4190688133239746,
+      "learning_rate": 0.0001930864197530864,
+      "loss": 3.6966,
+      "step": 568
+    },
+    {
+      "epoch": 1.2623405435385469,
+      "grad_norm": 2.1883962154388428,
+      "learning_rate": 0.0001928395061728395,
+      "loss": 3.607,
+      "step": 569
+    },
+    {
+      "epoch": 1.264559068219634,
+      "grad_norm": 2.2727067470550537,
+      "learning_rate": 0.00019259259259259257,
+      "loss": 3.5495,
+      "step": 570
+    },
+    {
+      "epoch": 1.266777592900721,
+      "grad_norm": 2.044931411743164,
+      "learning_rate": 0.00019234567901234566,
+      "loss": 3.3453,
+      "step": 571
+    },
+    {
+      "epoch": 1.268996117581808,
+      "grad_norm": 3.178436279296875,
+      "learning_rate": 0.00019209876543209873,
+      "loss": 3.6198,
+      "step": 572
+    },
+    {
+      "epoch": 1.2712146422628952,
+      "grad_norm": 2.7760982513427734,
+      "learning_rate": 0.00019185185185185182,
+      "loss": 3.621,
+      "step": 573
+    },
+    {
+      "epoch": 1.2734331669439822,
+      "grad_norm": 2.547375440597534,
+      "learning_rate": 0.00019160493827160495,
+      "loss": 3.604,
+      "step": 574
+    },
+    {
+      "epoch": 1.2756516916250693,
+      "grad_norm": 2.2375171184539795,
+      "learning_rate": 0.00019135802469135801,
+      "loss": 3.902,
+      "step": 575
+    },
+    {
+      "epoch": 1.2778702163061564,
+      "grad_norm": 2.2848920822143555,
+      "learning_rate": 0.0001911111111111111,
+      "loss": 3.2936,
+      "step": 576
+    },
+    {
+      "epoch": 1.2800887409872435,
+      "grad_norm": 2.776404857635498,
+      "learning_rate": 0.00019086419753086418,
+      "loss": 3.7362,
+      "step": 577
+    },
+    {
+      "epoch": 1.2823072656683305,
+      "grad_norm": 2.271362543106079,
+      "learning_rate": 0.00019061728395061727,
+      "loss": 3.814,
+      "step": 578
+    },
+    {
+      "epoch": 1.2845257903494176,
+      "grad_norm": 2.4814467430114746,
+      "learning_rate": 0.00019037037037037034,
+      "loss": 3.5541,
+      "step": 579
+    },
+    {
+      "epoch": 1.2867443150305047,
+      "grad_norm": 2.9616196155548096,
+      "learning_rate": 0.00019012345679012344,
+      "loss": 3.8433,
+      "step": 580
+    },
+    {
+      "epoch": 1.2889628397115918,
+      "grad_norm": 2.865147590637207,
+      "learning_rate": 0.00018987654320987653,
+      "loss": 3.9741,
+      "step": 581
+    },
+    {
+      "epoch": 1.2911813643926788,
+      "grad_norm": 2.4218595027923584,
+      "learning_rate": 0.0001896296296296296,
+      "loss": 3.674,
+      "step": 582
+    },
+    {
+      "epoch": 1.293399889073766,
+      "grad_norm": 2.4814767837524414,
+      "learning_rate": 0.00018938271604938272,
+      "loss": 3.2524,
+      "step": 583
+    },
+    {
+      "epoch": 1.295618413754853,
+      "grad_norm": 2.1578288078308105,
+      "learning_rate": 0.0001891358024691358,
+      "loss": 3.601,
+      "step": 584
+    },
+    {
+      "epoch": 1.29783693843594,
+      "grad_norm": 1.9944571256637573,
+      "learning_rate": 0.00018888888888888888,
+      "loss": 3.7919,
+      "step": 585
+    },
+    {
+      "epoch": 1.3000554631170271,
+      "grad_norm": 2.8170249462127686,
+      "learning_rate": 0.00018864197530864195,
+      "loss": 3.8147,
+      "step": 586
+    },
+    {
+      "epoch": 1.3022739877981142,
+      "grad_norm": 2.373061180114746,
+      "learning_rate": 0.00018839506172839505,
+      "loss": 3.7872,
+      "step": 587
+    },
+    {
+      "epoch": 1.3044925124792013,
+      "grad_norm": 2.30344557762146,
+      "learning_rate": 0.00018814814814814814,
+      "loss": 3.6505,
+      "step": 588
+    },
+    {
+      "epoch": 1.3067110371602884,
+      "grad_norm": 2.840280055999756,
+      "learning_rate": 0.0001879012345679012,
+      "loss": 3.8059,
+      "step": 589
+    },
+    {
+      "epoch": 1.3089295618413754,
+      "grad_norm": 1.8192697763442993,
+      "learning_rate": 0.0001876543209876543,
+      "loss": 3.5323,
+      "step": 590
+    },
+    {
+      "epoch": 1.3111480865224625,
+      "grad_norm": 2.1306910514831543,
+      "learning_rate": 0.00018740740740740737,
+      "loss": 3.5512,
+      "step": 591
+    },
+    {
+      "epoch": 1.3133666112035496,
+      "grad_norm": 2.4473209381103516,
+      "learning_rate": 0.0001871604938271605,
+      "loss": 3.5498,
+      "step": 592
+    },
+    {
+      "epoch": 1.3155851358846367,
+      "grad_norm": 2.419137477874756,
+      "learning_rate": 0.00018691358024691356,
+      "loss": 3.5344,
+      "step": 593
+    },
+    {
+      "epoch": 1.3178036605657237,
+      "grad_norm": 2.583573579788208,
+      "learning_rate": 0.00018666666666666666,
+      "loss": 3.9777,
+      "step": 594
+    },
+    {
+      "epoch": 1.3200221852468108,
+      "grad_norm": 2.5937602519989014,
+      "learning_rate": 0.00018641975308641973,
+      "loss": 3.8628,
+      "step": 595
+    },
+    {
+      "epoch": 1.3222407099278979,
+      "grad_norm": 2.763615369796753,
+      "learning_rate": 0.00018617283950617282,
+      "loss": 3.6152,
+      "step": 596
+    },
+    {
+      "epoch": 1.324459234608985,
+      "grad_norm": 2.8287930488586426,
+      "learning_rate": 0.00018592592592592592,
+      "loss": 3.4486,
+      "step": 597
+    },
+    {
+      "epoch": 1.326677759290072,
+      "grad_norm": 2.956371307373047,
+      "learning_rate": 0.00018567901234567899,
+      "loss": 3.5317,
+      "step": 598
+    },
+    {
+      "epoch": 1.328896283971159,
+      "grad_norm": 3.7075369358062744,
+      "learning_rate": 0.00018543209876543208,
+      "loss": 3.5077,
+      "step": 599
+    },
+    {
+      "epoch": 1.3311148086522462,
+      "grad_norm": 3.220616102218628,
+      "learning_rate": 0.00018518518518518515,
+      "loss": 3.6328,
+      "step": 600
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 2.72424578666687,
+      "learning_rate": 0.00018493827160493827,
+      "loss": 3.722,
+      "step": 601
+    },
+    {
+      "epoch": 1.3355518580144203,
+      "grad_norm": 2.402318239212036,
+      "learning_rate": 0.00018469135802469134,
+      "loss": 3.7801,
+      "step": 602
+    },
+    {
+      "epoch": 1.3377703826955074,
+      "grad_norm": 1.8593829870224,
+      "learning_rate": 0.00018444444444444443,
+      "loss": 3.45,
+      "step": 603
+    },
+    {
+      "epoch": 1.3399889073765945,
+      "grad_norm": 3.1333324909210205,
+      "learning_rate": 0.00018419753086419753,
+      "loss": 4.1691,
+      "step": 604
+    },
+    {
+      "epoch": 1.3422074320576816,
+      "grad_norm": 2.5560951232910156,
+      "learning_rate": 0.0001839506172839506,
+      "loss": 3.8546,
+      "step": 605
+    },
+    {
+      "epoch": 1.3444259567387689,
+      "grad_norm": 2.3715875148773193,
+      "learning_rate": 0.0001837037037037037,
+      "loss": 3.6814,
+      "step": 606
+    },
+    {
+      "epoch": 1.346644481419856,
+      "grad_norm": 1.8978787660598755,
+      "learning_rate": 0.00018345679012345676,
+      "loss": 3.815,
+      "step": 607
+    },
+    {
+      "epoch": 1.348863006100943,
+      "grad_norm": 1.8984980583190918,
+      "learning_rate": 0.00018320987654320986,
+      "loss": 3.4048,
+      "step": 608
+    },
+    {
+      "epoch": 1.35108153078203,
+      "grad_norm": 2.561610698699951,
+      "learning_rate": 0.00018296296296296292,
+      "loss": 3.845,
+      "step": 609
+    },
+    {
+      "epoch": 1.3533000554631172,
+      "grad_norm": 2.4611945152282715,
+      "learning_rate": 0.00018271604938271605,
+      "loss": 3.506,
+      "step": 610
+    },
+    {
+      "epoch": 1.3555185801442042,
+      "grad_norm": 2.7221291065216064,
+      "learning_rate": 0.00018246913580246914,
+      "loss": 3.6893,
+      "step": 611
+    },
+    {
+      "epoch": 1.3577371048252913,
+      "grad_norm": 3.262397050857544,
+      "learning_rate": 0.0001822222222222222,
+      "loss": 3.87,
+      "step": 612
+    },
+    {
+      "epoch": 1.3599556295063784,
+      "grad_norm": 2.528733491897583,
+      "learning_rate": 0.0001819753086419753,
+      "loss": 3.487,
+      "step": 613
+    },
+    {
+      "epoch": 1.3621741541874655,
+      "grad_norm": 2.3772521018981934,
+      "learning_rate": 0.00018172839506172837,
+      "loss": 3.5348,
+      "step": 614
+    },
+    {
+      "epoch": 1.3643926788685525,
+      "grad_norm": 2.9330554008483887,
+      "learning_rate": 0.00018148148148148147,
+      "loss": 3.7667,
+      "step": 615
+    },
+    {
+      "epoch": 1.3666112035496396,
+      "grad_norm": 2.550530195236206,
+      "learning_rate": 0.00018123456790123454,
+      "loss": 3.6363,
+      "step": 616
+    },
+    {
+      "epoch": 1.3688297282307267,
+      "grad_norm": 2.750720977783203,
+      "learning_rate": 0.00018098765432098763,
+      "loss": 3.8233,
+      "step": 617
+    },
+    {
+      "epoch": 1.3710482529118138,
+      "grad_norm": 2.5217208862304688,
+      "learning_rate": 0.00018074074074074075,
+      "loss": 3.9156,
+      "step": 618
+    },
+    {
+      "epoch": 1.3732667775929008,
+      "grad_norm": 2.152620553970337,
+      "learning_rate": 0.00018049382716049382,
+      "loss": 3.427,
+      "step": 619
+    },
+    {
+      "epoch": 1.375485302273988,
+      "grad_norm": 2.5827548503875732,
+      "learning_rate": 0.00018024691358024692,
+      "loss": 3.5684,
+      "step": 620
+    },
+    {
+      "epoch": 1.377703826955075,
+      "grad_norm": 2.5962836742401123,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 3.7622,
+      "step": 621
+    },
+    {
+      "epoch": 1.379922351636162,
+      "grad_norm": 2.4372382164001465,
+      "learning_rate": 0.00017975308641975308,
+      "loss": 3.5231,
+      "step": 622
+    },
+    {
+      "epoch": 1.3821408763172491,
+      "grad_norm": 2.7144031524658203,
+      "learning_rate": 0.00017950617283950615,
+      "loss": 3.6399,
+      "step": 623
+    },
+    {
+      "epoch": 1.3843594009983362,
+      "grad_norm": 2.639549970626831,
+      "learning_rate": 0.00017925925925925924,
+      "loss": 3.4395,
+      "step": 624
+    },
+    {
+      "epoch": 1.3865779256794233,
+      "grad_norm": 2.431058168411255,
+      "learning_rate": 0.0001790123456790123,
+      "loss": 3.5994,
+      "step": 625
+    },
+    {
+      "epoch": 1.3887964503605104,
+      "grad_norm": 2.3116865158081055,
+      "learning_rate": 0.0001787654320987654,
+      "loss": 3.3454,
+      "step": 626
+    },
+    {
+      "epoch": 1.3910149750415974,
+      "grad_norm": 2.7370753288269043,
+      "learning_rate": 0.00017851851851851853,
+      "loss": 3.2877,
+      "step": 627
+    },
+    {
+      "epoch": 1.3932334997226845,
+      "grad_norm": 2.3302462100982666,
+      "learning_rate": 0.0001782716049382716,
+      "loss": 3.4204,
+      "step": 628
+    },
+    {
+      "epoch": 1.3954520244037716,
+      "grad_norm": 2.5023751258850098,
+      "learning_rate": 0.0001780246913580247,
+      "loss": 3.3747,
+      "step": 629
+    },
+    {
+      "epoch": 1.3976705490848587,
+      "grad_norm": 2.4087181091308594,
+      "learning_rate": 0.00017777777777777776,
+      "loss": 3.6327,
+      "step": 630
+    },
+    {
+      "epoch": 1.3998890737659457,
+      "grad_norm": 2.511221170425415,
+      "learning_rate": 0.00017753086419753085,
+      "loss": 3.581,
+      "step": 631
+    },
+    {
+      "epoch": 1.4021075984470328,
+      "grad_norm": 3.098580837249756,
+      "learning_rate": 0.00017728395061728392,
+      "loss": 3.5801,
+      "step": 632
+    },
+    {
+      "epoch": 1.4043261231281199,
+      "grad_norm": 3.570006847381592,
+      "learning_rate": 0.00017703703703703702,
+      "loss": 3.6855,
+      "step": 633
+    },
+    {
+      "epoch": 1.406544647809207,
+      "grad_norm": 2.5198593139648438,
+      "learning_rate": 0.0001767901234567901,
+      "loss": 3.5308,
+      "step": 634
+    },
+    {
+      "epoch": 1.408763172490294,
+      "grad_norm": 2.291468858718872,
+      "learning_rate": 0.00017654320987654318,
+      "loss": 3.3816,
+      "step": 635
+    },
+    {
+      "epoch": 1.410981697171381,
+      "grad_norm": 2.405667304992676,
+      "learning_rate": 0.0001762962962962963,
+      "loss": 3.4956,
+      "step": 636
+    },
+    {
+      "epoch": 1.4132002218524682,
+      "grad_norm": 2.8232719898223877,
+      "learning_rate": 0.00017604938271604937,
+      "loss": 3.5802,
+      "step": 637
+    },
+    {
+      "epoch": 1.4154187465335553,
+      "grad_norm": 2.5068559646606445,
+      "learning_rate": 0.00017580246913580247,
+      "loss": 3.8268,
+      "step": 638
+    },
+    {
+      "epoch": 1.4176372712146423,
+      "grad_norm": 2.2451045513153076,
+      "learning_rate": 0.00017555555555555553,
+      "loss": 3.5048,
+      "step": 639
+    },
+    {
+      "epoch": 1.4198557958957294,
+      "grad_norm": 2.533236026763916,
+      "learning_rate": 0.00017530864197530863,
+      "loss": 3.5434,
+      "step": 640
+    },
+    {
+      "epoch": 1.4220743205768165,
+      "grad_norm": 2.2526965141296387,
+      "learning_rate": 0.00017506172839506172,
+      "loss": 3.7821,
+      "step": 641
+    },
+    {
+      "epoch": 1.4242928452579036,
+      "grad_norm": 2.5348198413848877,
+      "learning_rate": 0.0001748148148148148,
+      "loss": 3.5077,
+      "step": 642
+    },
+    {
+      "epoch": 1.4265113699389906,
+      "grad_norm": 2.55938458442688,
+      "learning_rate": 0.0001745679012345679,
+      "loss": 3.4471,
+      "step": 643
+    },
+    {
+      "epoch": 1.4287298946200777,
+      "grad_norm": 2.547743082046509,
+      "learning_rate": 0.00017432098765432096,
+      "loss": 3.8196,
+      "step": 644
+    },
+    {
+      "epoch": 1.4309484193011648,
+      "grad_norm": 2.5046439170837402,
+      "learning_rate": 0.00017407407407407408,
+      "loss": 3.6416,
+      "step": 645
+    },
+    {
+      "epoch": 1.4331669439822519,
+      "grad_norm": 2.5372109413146973,
+      "learning_rate": 0.00017382716049382715,
+      "loss": 3.5613,
+      "step": 646
+    },
+    {
+      "epoch": 1.435385468663339,
+      "grad_norm": 2.7168753147125244,
+      "learning_rate": 0.00017358024691358024,
+      "loss": 3.6951,
+      "step": 647
+    },
+    {
+      "epoch": 1.437603993344426,
+      "grad_norm": 2.5757460594177246,
+      "learning_rate": 0.0001733333333333333,
+      "loss": 3.4152,
+      "step": 648
+    },
+    {
+      "epoch": 1.439822518025513,
+      "grad_norm": 2.216088056564331,
+      "learning_rate": 0.0001730864197530864,
+      "loss": 3.7029,
+      "step": 649
+    },
+    {
+      "epoch": 1.4420410427066002,
+      "grad_norm": 2.3588953018188477,
+      "learning_rate": 0.0001728395061728395,
+      "loss": 3.5504,
+      "step": 650
+    },
+    {
+      "epoch": 1.4442595673876872,
+      "grad_norm": 2.413015604019165,
+      "learning_rate": 0.00017259259259259257,
+      "loss": 3.522,
+      "step": 651
+    },
+    {
+      "epoch": 1.4464780920687743,
+      "grad_norm": 2.4972872734069824,
+      "learning_rate": 0.00017234567901234566,
+      "loss": 3.4256,
+      "step": 652
+    },
+    {
+      "epoch": 1.4486966167498614,
+      "grad_norm": 2.972883939743042,
+      "learning_rate": 0.00017209876543209873,
+      "loss": 3.5276,
+      "step": 653
+    },
+    {
+      "epoch": 1.4509151414309485,
+      "grad_norm": 2.4234139919281006,
+      "learning_rate": 0.00017185185185185185,
+      "loss": 3.3781,
+      "step": 654
+    },
+    {
+      "epoch": 1.4531336661120355,
+      "grad_norm": 2.601092576980591,
+      "learning_rate": 0.00017160493827160492,
+      "loss": 3.3993,
+      "step": 655
+    },
+    {
+      "epoch": 1.4553521907931226,
+      "grad_norm": 2.804673194885254,
+      "learning_rate": 0.00017135802469135802,
+      "loss": 3.3073,
+      "step": 656
+    },
+    {
+      "epoch": 1.4575707154742097,
+      "grad_norm": 3.3741555213928223,
+      "learning_rate": 0.0001711111111111111,
+      "loss": 3.5957,
+      "step": 657
+    },
+    {
+      "epoch": 1.4597892401552968,
+      "grad_norm": 2.5167417526245117,
+      "learning_rate": 0.00017086419753086418,
+      "loss": 3.5033,
+      "step": 658
+    },
+    {
+      "epoch": 1.4620077648363838,
+      "grad_norm": 2.9775521755218506,
+      "learning_rate": 0.00017061728395061727,
+      "loss": 3.8389,
+      "step": 659
+    },
+    {
+      "epoch": 1.464226289517471,
+      "grad_norm": 2.811049222946167,
+      "learning_rate": 0.00017037037037037034,
+      "loss": 3.134,
+      "step": 660
+    },
+    {
+      "epoch": 1.466444814198558,
+      "grad_norm": 2.3152034282684326,
+      "learning_rate": 0.00017012345679012344,
+      "loss": 3.862,
+      "step": 661
+    },
+    {
+      "epoch": 1.468663338879645,
+      "grad_norm": 2.7881832122802734,
+      "learning_rate": 0.0001698765432098765,
+      "loss": 3.4658,
+      "step": 662
+    },
+    {
+      "epoch": 1.4708818635607321,
+      "grad_norm": 2.329094886779785,
+      "learning_rate": 0.00016962962962962963,
+      "loss": 3.7327,
+      "step": 663
+    },
+    {
+      "epoch": 1.4731003882418192,
+      "grad_norm": 2.7865848541259766,
+      "learning_rate": 0.00016938271604938272,
+      "loss": 3.446,
+      "step": 664
+    },
+    {
+      "epoch": 1.4753189129229063,
+      "grad_norm": 2.7451510429382324,
+      "learning_rate": 0.0001691358024691358,
+      "loss": 3.6779,
+      "step": 665
+    },
+    {
+      "epoch": 1.4775374376039934,
+      "grad_norm": 2.2423970699310303,
+      "learning_rate": 0.00016888888888888889,
+      "loss": 3.9731,
+      "step": 666
+    },
+    {
+      "epoch": 1.4797559622850804,
+      "grad_norm": 2.4603376388549805,
+      "learning_rate": 0.00016864197530864195,
+      "loss": 3.7756,
+      "step": 667
+    },
+    {
+      "epoch": 1.4819744869661675,
+      "grad_norm": 2.8876724243164062,
+      "learning_rate": 0.00016839506172839505,
+      "loss": 3.5894,
+      "step": 668
+    },
+    {
+      "epoch": 1.4841930116472546,
+      "grad_norm": 2.4098591804504395,
+      "learning_rate": 0.00016814814814814812,
+      "loss": 3.5765,
+      "step": 669
+    },
+    {
+      "epoch": 1.4864115363283417,
+      "grad_norm": 2.3790974617004395,
+      "learning_rate": 0.0001679012345679012,
+      "loss": 3.5637,
+      "step": 670
+    },
+    {
+      "epoch": 1.4886300610094287,
+      "grad_norm": 2.039487361907959,
+      "learning_rate": 0.00016765432098765433,
+      "loss": 3.4177,
+      "step": 671
+    },
+    {
+      "epoch": 1.4908485856905158,
+      "grad_norm": 2.15055251121521,
+      "learning_rate": 0.0001674074074074074,
+      "loss": 3.5661,
+      "step": 672
+    },
+    {
+      "epoch": 1.493067110371603,
+      "grad_norm": 2.2532403469085693,
+      "learning_rate": 0.0001671604938271605,
+      "loss": 3.5932,
+      "step": 673
+    },
+    {
+      "epoch": 1.49528563505269,
+      "grad_norm": 2.5822250843048096,
+      "learning_rate": 0.00016691358024691357,
+      "loss": 3.577,
+      "step": 674
+    },
+    {
+      "epoch": 1.497504159733777,
+      "grad_norm": 2.648679494857788,
+      "learning_rate": 0.00016666666666666666,
+      "loss": 3.5975,
+      "step": 675
+    },
+    {
+      "epoch": 1.4997226844148641,
+      "grad_norm": 3.0044548511505127,
+      "learning_rate": 0.00016641975308641973,
+      "loss": 3.6981,
+      "step": 676
+    },
+    {
+      "epoch": 1.5019412090959512,
+      "grad_norm": 2.3824853897094727,
+      "learning_rate": 0.00016617283950617282,
+      "loss": 3.603,
+      "step": 677
+    },
+    {
+      "epoch": 1.5041597337770383,
+      "grad_norm": 2.2382829189300537,
+      "learning_rate": 0.0001659259259259259,
+      "loss": 3.2749,
+      "step": 678
+    },
+    {
+      "epoch": 1.5063782584581253,
+      "grad_norm": 2.310223340988159,
+      "learning_rate": 0.000165679012345679,
+      "loss": 3.4763,
+      "step": 679
+    },
+    {
+      "epoch": 1.5085967831392124,
+      "grad_norm": 2.5231218338012695,
+      "learning_rate": 0.0001654320987654321,
+      "loss": 3.7748,
+      "step": 680
+    },
+    {
+      "epoch": 1.5108153078202995,
+      "grad_norm": 2.4063541889190674,
+      "learning_rate": 0.00016518518518518518,
+      "loss": 3.7067,
+      "step": 681
+    },
+    {
+      "epoch": 1.5130338325013866,
+      "grad_norm": 3.2637486457824707,
+      "learning_rate": 0.00016493827160493827,
+      "loss": 3.5397,
+      "step": 682
+    },
+    {
+      "epoch": 1.5152523571824736,
+      "grad_norm": 2.4993855953216553,
+      "learning_rate": 0.00016469135802469134,
+      "loss": 3.457,
+      "step": 683
+    },
+    {
+      "epoch": 1.5174708818635607,
+      "grad_norm": 2.356715440750122,
+      "learning_rate": 0.00016444444444444444,
+      "loss": 3.5776,
+      "step": 684
+    },
+    {
+      "epoch": 1.5196894065446478,
+      "grad_norm": 2.7361104488372803,
+      "learning_rate": 0.0001641975308641975,
+      "loss": 3.6191,
+      "step": 685
+    },
+    {
+      "epoch": 1.5219079312257349,
+      "grad_norm": 3.1861720085144043,
+      "learning_rate": 0.0001639506172839506,
+      "loss": 3.7852,
+      "step": 686
+    },
+    {
+      "epoch": 1.524126455906822,
+      "grad_norm": 2.447115659713745,
+      "learning_rate": 0.0001637037037037037,
+      "loss": 3.2529,
+      "step": 687
+    },
+    {
+      "epoch": 1.526344980587909,
+      "grad_norm": 2.6788201332092285,
+      "learning_rate": 0.00016345679012345676,
+      "loss": 3.7556,
+      "step": 688
+    },
+    {
+      "epoch": 1.528563505268996,
+      "grad_norm": 2.4821596145629883,
+      "learning_rate": 0.00016320987654320988,
+      "loss": 3.4607,
+      "step": 689
+    },
+    {
+      "epoch": 1.5307820299500832,
+      "grad_norm": 2.62007737159729,
+      "learning_rate": 0.00016296296296296295,
+      "loss": 3.6435,
+      "step": 690
+    },
+    {
+      "epoch": 1.5330005546311702,
+      "grad_norm": 2.634765625,
+      "learning_rate": 0.00016271604938271605,
+      "loss": 3.5597,
+      "step": 691
+    },
+    {
+      "epoch": 1.5352190793122573,
+      "grad_norm": 2.682448387145996,
+      "learning_rate": 0.00016246913580246912,
+      "loss": 3.8351,
+      "step": 692
+    },
+    {
+      "epoch": 1.5374376039933444,
+      "grad_norm": 2.7088520526885986,
+      "learning_rate": 0.0001622222222222222,
+      "loss": 3.7868,
+      "step": 693
+    },
+    {
+      "epoch": 1.5396561286744315,
+      "grad_norm": 2.7977161407470703,
+      "learning_rate": 0.0001619753086419753,
+      "loss": 3.4603,
+      "step": 694
+    },
+    {
+      "epoch": 1.5418746533555185,
+      "grad_norm": 2.480452299118042,
+      "learning_rate": 0.00016172839506172837,
+      "loss": 3.3448,
+      "step": 695
+    },
+    {
+      "epoch": 1.5440931780366056,
+      "grad_norm": 2.3126628398895264,
+      "learning_rate": 0.00016148148148148147,
+      "loss": 3.5302,
+      "step": 696
+    },
+    {
+      "epoch": 1.5463117027176927,
+      "grad_norm": 2.3544538021087646,
+      "learning_rate": 0.00016123456790123454,
+      "loss": 3.5133,
+      "step": 697
+    },
+    {
+      "epoch": 1.5485302273987798,
+      "grad_norm": 2.361126184463501,
+      "learning_rate": 0.00016098765432098766,
+      "loss": 3.4029,
+      "step": 698
+    },
+    {
+      "epoch": 1.5507487520798668,
+      "grad_norm": 2.3597121238708496,
+      "learning_rate": 0.00016074074074074073,
+      "loss": 3.5269,
+      "step": 699
+    },
+    {
+      "epoch": 1.552967276760954,
+      "grad_norm": 2.2022457122802734,
+      "learning_rate": 0.00016049382716049382,
+      "loss": 3.4527,
+      "step": 700
+    },
+    {
+      "epoch": 1.555185801442041,
+      "grad_norm": 2.4426145553588867,
+      "learning_rate": 0.00016024691358024692,
+      "loss": 3.3085,
+      "step": 701
+    },
+    {
+      "epoch": 1.557404326123128,
+      "grad_norm": 2.324388027191162,
+      "learning_rate": 0.00015999999999999999,
+      "loss": 3.4743,
+      "step": 702
+    },
+    {
+      "epoch": 1.5596228508042151,
+      "grad_norm": 2.6675150394439697,
+      "learning_rate": 0.00015975308641975308,
+      "loss": 3.9464,
+      "step": 703
+    },
+    {
+      "epoch": 1.5618413754853022,
+      "grad_norm": 2.441481590270996,
+      "learning_rate": 0.00015950617283950615,
+      "loss": 3.4983,
+      "step": 704
+    },
+    {
+      "epoch": 1.5640599001663893,
+      "grad_norm": 3.1443798542022705,
+      "learning_rate": 0.00015925925925925924,
+      "loss": 3.2591,
+      "step": 705
+    },
+    {
+      "epoch": 1.5662784248474764,
+      "grad_norm": 2.6116199493408203,
+      "learning_rate": 0.0001590123456790123,
+      "loss": 3.3392,
+      "step": 706
+    },
+    {
+      "epoch": 1.5684969495285634,
+      "grad_norm": 2.6220057010650635,
+      "learning_rate": 0.00015876543209876543,
+      "loss": 3.3352,
+      "step": 707
+    },
+    {
+      "epoch": 1.5707154742096505,
+      "grad_norm": 2.8911521434783936,
+      "learning_rate": 0.0001585185185185185,
+      "loss": 3.6233,
+      "step": 708
+    },
+    {
+      "epoch": 1.5729339988907376,
+      "grad_norm": 2.8349835872650146,
+      "learning_rate": 0.0001582716049382716,
+      "loss": 3.7228,
+      "step": 709
+    },
+    {
+      "epoch": 1.5751525235718247,
+      "grad_norm": 2.8813316822052,
+      "learning_rate": 0.0001580246913580247,
+      "loss": 3.6265,
+      "step": 710
+    },
+    {
+      "epoch": 1.5773710482529117,
+      "grad_norm": 2.6233460903167725,
+      "learning_rate": 0.00015777777777777776,
+      "loss": 3.7868,
+      "step": 711
+    },
+    {
+      "epoch": 1.5795895729339988,
+      "grad_norm": 2.634857416152954,
+      "learning_rate": 0.00015753086419753086,
+      "loss": 3.543,
+      "step": 712
+    },
+    {
+      "epoch": 1.581808097615086,
+      "grad_norm": 3.0219547748565674,
+      "learning_rate": 0.00015728395061728392,
+      "loss": 3.5713,
+      "step": 713
+    },
+    {
+      "epoch": 1.584026622296173,
+      "grad_norm": 2.318368673324585,
+      "learning_rate": 0.00015703703703703702,
+      "loss": 3.4954,
+      "step": 714
+    },
+    {
+      "epoch": 1.58624514697726,
+      "grad_norm": 2.4162113666534424,
+      "learning_rate": 0.0001567901234567901,
+      "loss": 3.5528,
+      "step": 715
+    },
+    {
+      "epoch": 1.5884636716583471,
+      "grad_norm": 2.4009268283843994,
+      "learning_rate": 0.0001565432098765432,
+      "loss": 3.6898,
+      "step": 716
+    },
+    {
+      "epoch": 1.5906821963394342,
+      "grad_norm": 2.5944018363952637,
+      "learning_rate": 0.0001562962962962963,
+      "loss": 3.6561,
+      "step": 717
+    },
+    {
+      "epoch": 1.5929007210205213,
+      "grad_norm": 3.3101136684417725,
+      "learning_rate": 0.00015604938271604937,
+      "loss": 3.6736,
+      "step": 718
+    },
+    {
+      "epoch": 1.5951192457016083,
+      "grad_norm": 2.729680299758911,
+      "learning_rate": 0.00015580246913580247,
+      "loss": 3.2612,
+      "step": 719
+    },
+    {
+      "epoch": 1.5973377703826954,
+      "grad_norm": 2.3222384452819824,
+      "learning_rate": 0.00015555555555555554,
+      "loss": 3.6561,
+      "step": 720
+    },
+    {
+      "epoch": 1.5995562950637825,
+      "grad_norm": 2.149651527404785,
+      "learning_rate": 0.00015530864197530863,
+      "loss": 3.5482,
+      "step": 721
+    },
+    {
+      "epoch": 1.6017748197448696,
+      "grad_norm": 2.397388458251953,
+      "learning_rate": 0.0001550617283950617,
+      "loss": 3.6164,
+      "step": 722
+    },
+    {
+      "epoch": 1.6039933444259566,
+      "grad_norm": 2.9447824954986572,
+      "learning_rate": 0.0001548148148148148,
+      "loss": 3.1427,
+      "step": 723
+    },
+    {
+      "epoch": 1.6062118691070437,
+      "grad_norm": 3.0858662128448486,
+      "learning_rate": 0.00015456790123456792,
+      "loss": 3.6733,
+      "step": 724
+    },
+    {
+      "epoch": 1.6084303937881308,
+      "grad_norm": 2.5016748905181885,
+      "learning_rate": 0.00015432098765432098,
+      "loss": 3.3463,
+      "step": 725
+    },
+    {
+      "epoch": 1.6106489184692179,
+      "grad_norm": 2.91471266746521,
+      "learning_rate": 0.00015407407407407408,
+      "loss": 3.6776,
+      "step": 726
+    },
+    {
+      "epoch": 1.612867443150305,
+      "grad_norm": 2.4191195964813232,
+      "learning_rate": 0.00015382716049382715,
+      "loss": 3.2588,
+      "step": 727
+    },
+    {
+      "epoch": 1.615085967831392,
+      "grad_norm": 3.139770030975342,
+      "learning_rate": 0.00015358024691358024,
+      "loss": 3.562,
+      "step": 728
+    },
+    {
+      "epoch": 1.617304492512479,
+      "grad_norm": 2.432326316833496,
+      "learning_rate": 0.0001533333333333333,
+      "loss": 3.776,
+      "step": 729
+    },
+    {
+      "epoch": 1.6195230171935662,
+      "grad_norm": 2.3106398582458496,
+      "learning_rate": 0.0001530864197530864,
+      "loss": 3.3693,
+      "step": 730
+    },
+    {
+      "epoch": 1.6217415418746532,
+      "grad_norm": 2.363118886947632,
+      "learning_rate": 0.00015283950617283947,
+      "loss": 3.5083,
+      "step": 731
+    },
+    {
+      "epoch": 1.6239600665557403,
+      "grad_norm": 2.5189080238342285,
+      "learning_rate": 0.00015259259259259257,
+      "loss": 3.2401,
+      "step": 732
+    },
+    {
+      "epoch": 1.6261785912368274,
+      "grad_norm": 2.1984434127807617,
+      "learning_rate": 0.0001523456790123457,
+      "loss": 3.2033,
+      "step": 733
+    },
+    {
+      "epoch": 1.6283971159179145,
+      "grad_norm": 3.3136227130889893,
+      "learning_rate": 0.00015209876543209873,
+      "loss": 3.4447,
+      "step": 734
+    },
+    {
+      "epoch": 1.6306156405990015,
+      "grad_norm": 2.3485517501831055,
+      "learning_rate": 0.00015185185185185185,
+      "loss": 3.6353,
+      "step": 735
+    },
+    {
+      "epoch": 1.6328341652800886,
+      "grad_norm": 2.5111780166625977,
+      "learning_rate": 0.00015160493827160492,
+      "loss": 3.48,
+      "step": 736
+    },
+    {
+      "epoch": 1.6350526899611757,
+      "grad_norm": 2.6205902099609375,
+      "learning_rate": 0.00015135802469135802,
+      "loss": 3.3098,
+      "step": 737
+    },
+    {
+      "epoch": 1.6372712146422628,
+      "grad_norm": 2.4828739166259766,
+      "learning_rate": 0.00015111111111111109,
+      "loss": 3.1632,
+      "step": 738
+    },
+    {
+      "epoch": 1.6394897393233498,
+      "grad_norm": 2.4862096309661865,
+      "learning_rate": 0.00015086419753086418,
+      "loss": 3.5605,
+      "step": 739
+    },
+    {
+      "epoch": 1.641708264004437,
+      "grad_norm": 2.6750941276550293,
+      "learning_rate": 0.00015061728395061728,
+      "loss": 3.671,
+      "step": 740
+    },
+    {
+      "epoch": 1.643926788685524,
+      "grad_norm": 2.62343168258667,
+      "learning_rate": 0.00015037037037037034,
+      "loss": 3.7188,
+      "step": 741
+    },
+    {
+      "epoch": 1.646145313366611,
+      "grad_norm": 2.8903000354766846,
+      "learning_rate": 0.00015012345679012347,
+      "loss": 3.537,
+      "step": 742
+    },
+    {
+      "epoch": 1.6483638380476981,
+      "grad_norm": 2.331423759460449,
+      "learning_rate": 0.00014987654320987653,
+      "loss": 3.6107,
+      "step": 743
+    },
+    {
+      "epoch": 1.6505823627287852,
+      "grad_norm": 3.193272829055786,
+      "learning_rate": 0.00014962962962962963,
+      "loss": 3.7866,
+      "step": 744
+    },
+    {
+      "epoch": 1.6528008874098723,
+      "grad_norm": 2.448875665664673,
+      "learning_rate": 0.0001493827160493827,
+      "loss": 3.4041,
+      "step": 745
+    },
+    {
+      "epoch": 1.6550194120909594,
+      "grad_norm": 2.88864803314209,
+      "learning_rate": 0.0001491358024691358,
+      "loss": 3.7387,
+      "step": 746
+    },
+    {
+      "epoch": 1.6572379367720464,
+      "grad_norm": 3.1451332569122314,
+      "learning_rate": 0.00014888888888888886,
+      "loss": 3.6896,
+      "step": 747
+    },
+    {
+      "epoch": 1.6594564614531335,
+      "grad_norm": 2.532987356185913,
+      "learning_rate": 0.00014864197530864196,
+      "loss": 3.4974,
+      "step": 748
+    },
+    {
+      "epoch": 1.6616749861342206,
+      "grad_norm": 2.3754332065582275,
+      "learning_rate": 0.00014839506172839505,
+      "loss": 3.5436,
+      "step": 749
+    },
+    {
+      "epoch": 1.6638935108153077,
+      "grad_norm": 2.959876775741577,
+      "learning_rate": 0.00014814814814814812,
+      "loss": 3.4029,
+      "step": 750
+    },
+    {
+      "epoch": 1.6661120354963947,
+      "grad_norm": 2.6524274349212646,
+      "learning_rate": 0.00014790123456790124,
+      "loss": 3.4003,
+      "step": 751
+    },
+    {
+      "epoch": 1.6683305601774818,
+      "grad_norm": 2.3888120651245117,
+      "learning_rate": 0.0001476543209876543,
+      "loss": 3.3308,
+      "step": 752
+    },
+    {
+      "epoch": 1.670549084858569,
+      "grad_norm": 2.916497230529785,
+      "learning_rate": 0.0001474074074074074,
+      "loss": 3.4731,
+      "step": 753
+    },
+    {
+      "epoch": 1.672767609539656,
+      "grad_norm": 3.058265447616577,
+      "learning_rate": 0.00014716049382716047,
+      "loss": 3.6551,
+      "step": 754
+    },
+    {
+      "epoch": 1.674986134220743,
+      "grad_norm": 3.679319381713867,
+      "learning_rate": 0.00014691358024691357,
+      "loss": 3.8068,
+      "step": 755
+    },
+    {
+      "epoch": 1.6772046589018301,
+      "grad_norm": 2.996368169784546,
+      "learning_rate": 0.00014666666666666664,
+      "loss": 3.8401,
+      "step": 756
+    },
+    {
+      "epoch": 1.6794231835829172,
+      "grad_norm": 2.471221685409546,
+      "learning_rate": 0.00014641975308641973,
+      "loss": 3.4906,
+      "step": 757
+    },
+    {
+      "epoch": 1.6816417082640043,
+      "grad_norm": 2.5287835597991943,
+      "learning_rate": 0.00014617283950617283,
+      "loss": 3.5408,
+      "step": 758
+    },
+    {
+      "epoch": 1.6838602329450914,
+      "grad_norm": 2.781367778778076,
+      "learning_rate": 0.00014592592592592592,
+      "loss": 3.0299,
+      "step": 759
+    },
+    {
+      "epoch": 1.6860787576261784,
+      "grad_norm": 2.7500503063201904,
+      "learning_rate": 0.000145679012345679,
+      "loss": 3.3852,
+      "step": 760
+    },
+    {
+      "epoch": 1.6882972823072657,
+      "grad_norm": 2.4972012042999268,
+      "learning_rate": 0.00014543209876543208,
+      "loss": 3.4919,
+      "step": 761
+    },
+    {
+      "epoch": 1.6905158069883528,
+      "grad_norm": 2.5957489013671875,
+      "learning_rate": 0.00014518518518518518,
+      "loss": 3.3049,
+      "step": 762
+    },
+    {
+      "epoch": 1.6927343316694399,
+      "grad_norm": 3.2520978450775146,
+      "learning_rate": 0.00014493827160493825,
+      "loss": 3.7281,
+      "step": 763
+    },
+    {
+      "epoch": 1.694952856350527,
+      "grad_norm": 2.4658021926879883,
+      "learning_rate": 0.00014469135802469134,
+      "loss": 3.4663,
+      "step": 764
+    },
+    {
+      "epoch": 1.697171381031614,
+      "grad_norm": 2.8267886638641357,
+      "learning_rate": 0.0001444444444444444,
+      "loss": 3.7039,
+      "step": 765
+    },
+    {
+      "epoch": 1.699389905712701,
+      "grad_norm": 2.4844677448272705,
+      "learning_rate": 0.00014419753086419753,
+      "loss": 3.3323,
+      "step": 766
+    },
+    {
+      "epoch": 1.7016084303937882,
+      "grad_norm": 2.4565207958221436,
+      "learning_rate": 0.0001439506172839506,
+      "loss": 3.6953,
+      "step": 767
+    },
+    {
+      "epoch": 1.7038269550748752,
+      "grad_norm": 2.596930742263794,
+      "learning_rate": 0.0001437037037037037,
+      "loss": 3.4619,
+      "step": 768
+    },
+    {
+      "epoch": 1.7060454797559623,
+      "grad_norm": 2.167240619659424,
+      "learning_rate": 0.00014345679012345676,
+      "loss": 3.2668,
+      "step": 769
+    },
+    {
+      "epoch": 1.7082640044370494,
+      "grad_norm": 2.82932186126709,
+      "learning_rate": 0.00014320987654320986,
+      "loss": 3.5777,
+      "step": 770
+    },
+    {
+      "epoch": 1.7104825291181365,
+      "grad_norm": 2.2116572856903076,
+      "learning_rate": 0.00014296296296296295,
+      "loss": 3.5778,
+      "step": 771
+    },
+    {
+      "epoch": 1.7127010537992235,
+      "grad_norm": 2.6566524505615234,
+      "learning_rate": 0.00014271604938271602,
+      "loss": 3.4544,
+      "step": 772
+    },
+    {
+      "epoch": 1.7149195784803106,
+      "grad_norm": 2.587185859680176,
+      "learning_rate": 0.00014246913580246912,
+      "loss": 3.6396,
+      "step": 773
+    },
+    {
+      "epoch": 1.7171381031613977,
+      "grad_norm": 2.500413656234741,
+      "learning_rate": 0.0001422222222222222,
+      "loss": 3.2764,
+      "step": 774
+    },
+    {
+      "epoch": 1.7193566278424848,
+      "grad_norm": 2.5445940494537354,
+      "learning_rate": 0.0001419753086419753,
+      "loss": 3.6026,
+      "step": 775
+    },
+    {
+      "epoch": 1.7215751525235718,
+      "grad_norm": 2.389737129211426,
+      "learning_rate": 0.00014172839506172838,
+      "loss": 3.499,
+      "step": 776
+    },
+    {
+      "epoch": 1.723793677204659,
+      "grad_norm": 2.0811398029327393,
+      "learning_rate": 0.00014148148148148147,
+      "loss": 3.4701,
+      "step": 777
+    },
+    {
+      "epoch": 1.726012201885746,
+      "grad_norm": 3.3282265663146973,
+      "learning_rate": 0.00014123456790123454,
+      "loss": 3.627,
+      "step": 778
+    },
+    {
+      "epoch": 1.728230726566833,
+      "grad_norm": 2.3815293312072754,
+      "learning_rate": 0.00014098765432098763,
+      "loss": 3.8144,
+      "step": 779
+    },
+    {
+      "epoch": 1.7304492512479202,
+      "grad_norm": 2.60975980758667,
+      "learning_rate": 0.00014074074074074073,
+      "loss": 3.6556,
+      "step": 780
+    },
+    {
+      "epoch": 1.7326677759290072,
+      "grad_norm": 2.5978195667266846,
+      "learning_rate": 0.00014049382716049382,
+      "loss": 3.279,
+      "step": 781
+    },
+    {
+      "epoch": 1.7348863006100943,
+      "grad_norm": 3.0012166500091553,
+      "learning_rate": 0.0001402469135802469,
+      "loss": 3.3359,
+      "step": 782
+    },
+    {
+      "epoch": 1.7371048252911814,
+      "grad_norm": 3.0486395359039307,
+      "learning_rate": 0.00014,
+      "loss": 3.4418,
+      "step": 783
+    },
+    {
+      "epoch": 1.7393233499722685,
+      "grad_norm": 3.04864239692688,
+      "learning_rate": 0.00013975308641975308,
+      "loss": 3.4808,
+      "step": 784
+    },
+    {
+      "epoch": 1.7415418746533555,
+      "grad_norm": 2.458759307861328,
+      "learning_rate": 0.00013950617283950615,
+      "loss": 3.4882,
+      "step": 785
+    },
+    {
+      "epoch": 1.7437603993344426,
+      "grad_norm": 2.859830617904663,
+      "learning_rate": 0.00013925925925925925,
+      "loss": 3.3221,
+      "step": 786
+    },
+    {
+      "epoch": 1.7459789240155297,
+      "grad_norm": 2.521199941635132,
+      "learning_rate": 0.00013901234567901231,
+      "loss": 3.2116,
+      "step": 787
+    },
+    {
+      "epoch": 1.7481974486966168,
+      "grad_norm": 3.0465927124023438,
+      "learning_rate": 0.00013876543209876544,
+      "loss": 3.6086,
+      "step": 788
+    },
+    {
+      "epoch": 1.7504159733777038,
+      "grad_norm": 2.9240145683288574,
+      "learning_rate": 0.0001385185185185185,
+      "loss": 3.2652,
+      "step": 789
+    },
+    {
+      "epoch": 1.752634498058791,
+      "grad_norm": 2.173752784729004,
+      "learning_rate": 0.0001382716049382716,
+      "loss": 3.2935,
+      "step": 790
+    },
+    {
+      "epoch": 1.754853022739878,
+      "grad_norm": 2.4119791984558105,
+      "learning_rate": 0.00013802469135802467,
+      "loss": 3.4755,
+      "step": 791
+    },
+    {
+      "epoch": 1.757071547420965,
+      "grad_norm": 2.809386730194092,
+      "learning_rate": 0.00013777777777777776,
+      "loss": 3.447,
+      "step": 792
+    },
+    {
+      "epoch": 1.7592900721020521,
+      "grad_norm": 2.4718735218048096,
+      "learning_rate": 0.00013753086419753086,
+      "loss": 3.3193,
+      "step": 793
+    },
+    {
+      "epoch": 1.7615085967831392,
+      "grad_norm": 2.354329824447632,
+      "learning_rate": 0.00013728395061728393,
+      "loss": 3.4677,
+      "step": 794
+    },
+    {
+      "epoch": 1.7637271214642263,
+      "grad_norm": 2.702268362045288,
+      "learning_rate": 0.00013703703703703702,
+      "loss": 3.4354,
+      "step": 795
+    },
+    {
+      "epoch": 1.7659456461453134,
+      "grad_norm": 3.288127899169922,
+      "learning_rate": 0.00013679012345679012,
+      "loss": 3.5244,
+      "step": 796
+    },
+    {
+      "epoch": 1.7681641708264004,
+      "grad_norm": 2.871826171875,
+      "learning_rate": 0.0001365432098765432,
+      "loss": 3.6426,
+      "step": 797
+    },
+    {
+      "epoch": 1.7703826955074875,
+      "grad_norm": 2.7024428844451904,
+      "learning_rate": 0.00013629629629629628,
+      "loss": 3.5829,
+      "step": 798
+    },
+    {
+      "epoch": 1.7726012201885746,
+      "grad_norm": 2.368344783782959,
+      "learning_rate": 0.00013604938271604937,
+      "loss": 3.3395,
+      "step": 799
+    },
+    {
+      "epoch": 1.7748197448696617,
+      "grad_norm": 2.1372947692871094,
+      "learning_rate": 0.00013580246913580244,
+      "loss": 3.313,
+      "step": 800
+    },
+    {
+      "epoch": 1.7770382695507487,
+      "grad_norm": 2.5716147422790527,
+      "learning_rate": 0.00013555555555555554,
+      "loss": 3.9186,
+      "step": 801
+    },
+    {
+      "epoch": 1.7792567942318358,
+      "grad_norm": 3.0155889987945557,
+      "learning_rate": 0.00013530864197530863,
+      "loss": 3.5561,
+      "step": 802
+    },
+    {
+      "epoch": 1.7814753189129229,
+      "grad_norm": 2.9205029010772705,
+      "learning_rate": 0.00013506172839506173,
+      "loss": 3.5541,
+      "step": 803
+    },
+    {
+      "epoch": 1.78369384359401,
+      "grad_norm": 4.023229598999023,
+      "learning_rate": 0.0001348148148148148,
+      "loss": 3.4834,
+      "step": 804
+    },
+    {
+      "epoch": 1.785912368275097,
+      "grad_norm": 2.9994752407073975,
+      "learning_rate": 0.0001345679012345679,
+      "loss": 3.4544,
+      "step": 805
+    },
+    {
+      "epoch": 1.788130892956184,
+      "grad_norm": 2.325016736984253,
+      "learning_rate": 0.00013432098765432099,
+      "loss": 3.3835,
+      "step": 806
+    },
+    {
+      "epoch": 1.7903494176372712,
+      "grad_norm": 3.1206588745117188,
+      "learning_rate": 0.00013407407407407405,
+      "loss": 3.3547,
+      "step": 807
+    },
+    {
+      "epoch": 1.7925679423183583,
+      "grad_norm": 2.3782248497009277,
+      "learning_rate": 0.00013382716049382715,
+      "loss": 3.3987,
+      "step": 808
+    },
+    {
+      "epoch": 1.7947864669994453,
+      "grad_norm": 2.56799054145813,
+      "learning_rate": 0.00013358024691358022,
+      "loss": 3.2373,
+      "step": 809
+    },
+    {
+      "epoch": 1.7970049916805324,
+      "grad_norm": 3.3565545082092285,
+      "learning_rate": 0.0001333333333333333,
+      "loss": 3.5856,
+      "step": 810
+    },
+    {
+      "epoch": 1.7992235163616195,
+      "grad_norm": 2.997497797012329,
+      "learning_rate": 0.0001330864197530864,
+      "loss": 3.6851,
+      "step": 811
+    },
+    {
+      "epoch": 1.8014420410427066,
+      "grad_norm": 3.2373688220977783,
+      "learning_rate": 0.0001328395061728395,
+      "loss": 3.4384,
+      "step": 812
+    },
+    {
+      "epoch": 1.8036605657237936,
+      "grad_norm": 2.473341941833496,
+      "learning_rate": 0.00013259259259259257,
+      "loss": 3.3949,
+      "step": 813
+    },
+    {
+      "epoch": 1.8058790904048807,
+      "grad_norm": 2.422246217727661,
+      "learning_rate": 0.00013234567901234567,
+      "loss": 3.3475,
+      "step": 814
+    },
+    {
+      "epoch": 1.8080976150859678,
+      "grad_norm": 2.4680254459381104,
+      "learning_rate": 0.00013209876543209876,
+      "loss": 3.3584,
+      "step": 815
+    },
+    {
+      "epoch": 1.8103161397670549,
+      "grad_norm": 2.4544200897216797,
+      "learning_rate": 0.00013185185185185183,
+      "loss": 3.7715,
+      "step": 816
+    },
+    {
+      "epoch": 1.8125346644481422,
+      "grad_norm": 2.6361942291259766,
+      "learning_rate": 0.00013160493827160492,
+      "loss": 3.4648,
+      "step": 817
+    },
+    {
+      "epoch": 1.8147531891292292,
+      "grad_norm": 2.551222562789917,
+      "learning_rate": 0.00013135802469135802,
+      "loss": 3.3455,
+      "step": 818
+    },
+    {
+      "epoch": 1.8169717138103163,
+      "grad_norm": 2.38480806350708,
+      "learning_rate": 0.00013111111111111111,
+      "loss": 3.1088,
+      "step": 819
+    },
+    {
+      "epoch": 1.8191902384914034,
+      "grad_norm": 2.5755064487457275,
+      "learning_rate": 0.00013086419753086418,
+      "loss": 3.3683,
+      "step": 820
+    },
+    {
+      "epoch": 1.8214087631724905,
+      "grad_norm": 2.2944252490997314,
+      "learning_rate": 0.00013061728395061728,
+      "loss": 3.4129,
+      "step": 821
+    },
+    {
+      "epoch": 1.8236272878535775,
+      "grad_norm": 2.639906644821167,
+      "learning_rate": 0.00013037037037037035,
+      "loss": 3.329,
+      "step": 822
+    },
+    {
+      "epoch": 1.8258458125346646,
+      "grad_norm": 2.9540956020355225,
+      "learning_rate": 0.00013012345679012344,
+      "loss": 3.5146,
+      "step": 823
+    },
+    {
+      "epoch": 1.8280643372157517,
+      "grad_norm": 2.600825309753418,
+      "learning_rate": 0.00012987654320987654,
+      "loss": 3.2789,
+      "step": 824
+    },
+    {
+      "epoch": 1.8302828618968388,
+      "grad_norm": 2.7420971393585205,
+      "learning_rate": 0.0001296296296296296,
+      "loss": 3.4632,
+      "step": 825
+    },
+    {
+      "epoch": 1.8325013865779258,
+      "grad_norm": 2.7160956859588623,
+      "learning_rate": 0.0001293827160493827,
+      "loss": 3.3109,
+      "step": 826
+    },
+    {
+      "epoch": 1.834719911259013,
+      "grad_norm": 2.7182745933532715,
+      "learning_rate": 0.0001291358024691358,
+      "loss": 3.7874,
+      "step": 827
+    },
+    {
+      "epoch": 1.8369384359401,
+      "grad_norm": 2.429978847503662,
+      "learning_rate": 0.0001288888888888889,
+      "loss": 3.3848,
+      "step": 828
+    },
+    {
+      "epoch": 1.839156960621187,
+      "grad_norm": 3.383929491043091,
+      "learning_rate": 0.00012864197530864196,
+      "loss": 3.4906,
+      "step": 829
+    },
+    {
+      "epoch": 1.8413754853022741,
+      "grad_norm": 2.9490764141082764,
+      "learning_rate": 0.00012839506172839505,
+      "loss": 3.2072,
+      "step": 830
+    },
+    {
+      "epoch": 1.8435940099833612,
+      "grad_norm": 2.930039405822754,
+      "learning_rate": 0.00012814814814814812,
+      "loss": 3.5364,
+      "step": 831
+    },
+    {
+      "epoch": 1.8458125346644483,
+      "grad_norm": 2.93581485748291,
+      "learning_rate": 0.00012790123456790122,
+      "loss": 3.678,
+      "step": 832
+    },
+    {
+      "epoch": 1.8480310593455354,
+      "grad_norm": 2.948397397994995,
+      "learning_rate": 0.0001276543209876543,
+      "loss": 3.3107,
+      "step": 833
+    },
+    {
+      "epoch": 1.8502495840266224,
+      "grad_norm": 2.817746162414551,
+      "learning_rate": 0.0001274074074074074,
+      "loss": 3.4328,
+      "step": 834
+    },
+    {
+      "epoch": 1.8524681087077095,
+      "grad_norm": 2.260610580444336,
+      "learning_rate": 0.00012716049382716047,
+      "loss": 3.226,
+      "step": 835
+    },
+    {
+      "epoch": 1.8546866333887966,
+      "grad_norm": 2.4394407272338867,
+      "learning_rate": 0.00012691358024691357,
+      "loss": 3.6892,
+      "step": 836
+    },
+    {
+      "epoch": 1.8569051580698837,
+      "grad_norm": 2.5678303241729736,
+      "learning_rate": 0.00012666666666666666,
+      "loss": 3.3997,
+      "step": 837
+    },
+    {
+      "epoch": 1.8591236827509707,
+      "grad_norm": 2.6319210529327393,
+      "learning_rate": 0.00012641975308641973,
+      "loss": 3.2399,
+      "step": 838
+    },
+    {
+      "epoch": 1.8613422074320578,
+      "grad_norm": 3.285332679748535,
+      "learning_rate": 0.00012617283950617283,
+      "loss": 3.2387,
+      "step": 839
+    },
+    {
+      "epoch": 1.8635607321131449,
+      "grad_norm": 2.860302686691284,
+      "learning_rate": 0.0001259259259259259,
+      "loss": 3.3642,
+      "step": 840
+    },
+    {
+      "epoch": 1.865779256794232,
+      "grad_norm": 2.9101974964141846,
+      "learning_rate": 0.00012567901234567902,
+      "loss": 3.4953,
+      "step": 841
+    },
+    {
+      "epoch": 1.867997781475319,
+      "grad_norm": 2.4221808910369873,
+      "learning_rate": 0.00012543209876543209,
+      "loss": 3.3996,
+      "step": 842
+    },
+    {
+      "epoch": 1.870216306156406,
+      "grad_norm": 2.6599228382110596,
+      "learning_rate": 0.00012518518518518518,
+      "loss": 3.3064,
+      "step": 843
+    },
+    {
+      "epoch": 1.8724348308374932,
+      "grad_norm": 3.2023165225982666,
+      "learning_rate": 0.00012493827160493825,
+      "loss": 3.6134,
+      "step": 844
+    },
+    {
+      "epoch": 1.8746533555185803,
+      "grad_norm": 3.2467570304870605,
+      "learning_rate": 0.00012469135802469134,
+      "loss": 3.7816,
+      "step": 845
+    },
+    {
+      "epoch": 1.8768718801996673,
+      "grad_norm": 2.513582944869995,
+      "learning_rate": 0.00012444444444444444,
+      "loss": 3.4385,
+      "step": 846
+    },
+    {
+      "epoch": 1.8790904048807544,
+      "grad_norm": 2.5699844360351562,
+      "learning_rate": 0.0001241975308641975,
+      "loss": 3.4459,
+      "step": 847
+    },
+    {
+      "epoch": 1.8813089295618415,
+      "grad_norm": 3.0309741497039795,
+      "learning_rate": 0.0001239506172839506,
+      "loss": 3.1851,
+      "step": 848
+    },
+    {
+      "epoch": 1.8835274542429286,
+      "grad_norm": 2.1542470455169678,
+      "learning_rate": 0.0001237037037037037,
+      "loss": 3.504,
+      "step": 849
+    },
+    {
+      "epoch": 1.8857459789240156,
+      "grad_norm": 2.4568450450897217,
+      "learning_rate": 0.0001234567901234568,
+      "loss": 3.6159,
+      "step": 850
+    },
+    {
+      "epoch": 1.8879645036051027,
+      "grad_norm": 2.9098117351531982,
+      "learning_rate": 0.00012320987654320986,
+      "loss": 3.4748,
+      "step": 851
+    },
+    {
+      "epoch": 1.8901830282861898,
+      "grad_norm": 2.754383087158203,
+      "learning_rate": 0.00012296296296296296,
+      "loss": 3.3096,
+      "step": 852
+    },
+    {
+      "epoch": 1.8924015529672769,
+      "grad_norm": 3.2100822925567627,
+      "learning_rate": 0.00012271604938271602,
+      "loss": 3.5261,
+      "step": 853
+    },
+    {
+      "epoch": 1.894620077648364,
+      "grad_norm": 2.5671546459198,
+      "learning_rate": 0.00012246913580246912,
+      "loss": 3.5547,
+      "step": 854
+    },
+    {
+      "epoch": 1.896838602329451,
+      "grad_norm": 2.6438872814178467,
+      "learning_rate": 0.00012222222222222221,
+      "loss": 3.6573,
+      "step": 855
+    },
+    {
+      "epoch": 1.899057127010538,
+      "grad_norm": 2.141934633255005,
+      "learning_rate": 0.00012197530864197531,
+      "loss": 3.3807,
+      "step": 856
+    },
+    {
+      "epoch": 1.9012756516916252,
+      "grad_norm": 2.698051691055298,
+      "learning_rate": 0.00012172839506172839,
+      "loss": 3.216,
+      "step": 857
+    },
+    {
+      "epoch": 1.9034941763727122,
+      "grad_norm": 2.967970371246338,
+      "learning_rate": 0.00012148148148148147,
+      "loss": 3.361,
+      "step": 858
+    },
+    {
+      "epoch": 1.9057127010537993,
+      "grad_norm": 2.7769055366516113,
+      "learning_rate": 0.00012123456790123455,
+      "loss": 3.5638,
+      "step": 859
+    },
+    {
+      "epoch": 1.9079312257348864,
+      "grad_norm": 2.5321719646453857,
+      "learning_rate": 0.00012098765432098764,
+      "loss": 3.3964,
+      "step": 860
+    },
+    {
+      "epoch": 1.9101497504159735,
+      "grad_norm": 2.4845712184906006,
+      "learning_rate": 0.00012074074074074073,
+      "loss": 3.7885,
+      "step": 861
+    },
+    {
+      "epoch": 1.9123682750970605,
+      "grad_norm": 2.9583818912506104,
+      "learning_rate": 0.00012049382716049381,
+      "loss": 3.1101,
+      "step": 862
+    },
+    {
+      "epoch": 1.9145867997781476,
+      "grad_norm": 2.561662435531616,
+      "learning_rate": 0.0001202469135802469,
+      "loss": 3.0722,
+      "step": 863
+    },
+    {
+      "epoch": 1.9168053244592347,
+      "grad_norm": 3.3608789443969727,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 3.2966,
+      "step": 864
+    },
+    {
+      "epoch": 1.9190238491403218,
+      "grad_norm": 3.1011741161346436,
+      "learning_rate": 0.00011975308641975308,
+      "loss": 3.5144,
+      "step": 865
+    },
+    {
+      "epoch": 1.9212423738214088,
+      "grad_norm": 3.8238799571990967,
+      "learning_rate": 0.00011950617283950617,
+      "loss": 3.6266,
+      "step": 866
+    },
+    {
+      "epoch": 1.923460898502496,
+      "grad_norm": 3.0378761291503906,
+      "learning_rate": 0.00011925925925925925,
+      "loss": 3.1787,
+      "step": 867
+    },
+    {
+      "epoch": 1.925679423183583,
+      "grad_norm": 3.342501163482666,
+      "learning_rate": 0.00011901234567901233,
+      "loss": 3.8002,
+      "step": 868
+    },
+    {
+      "epoch": 1.92789794786467,
+      "grad_norm": 3.0960371494293213,
+      "learning_rate": 0.00011876543209876541,
+      "loss": 3.6183,
+      "step": 869
+    },
+    {
+      "epoch": 1.9301164725457571,
+      "grad_norm": 3.176086902618408,
+      "learning_rate": 0.0001185185185185185,
+      "loss": 3.6755,
+      "step": 870
+    },
+    {
+      "epoch": 1.9323349972268442,
+      "grad_norm": 2.921318531036377,
+      "learning_rate": 0.0001182716049382716,
+      "loss": 3.5097,
+      "step": 871
+    },
+    {
+      "epoch": 1.9345535219079313,
+      "grad_norm": 3.1497366428375244,
+      "learning_rate": 0.00011802469135802468,
+      "loss": 3.4124,
+      "step": 872
+    },
+    {
+      "epoch": 1.9367720465890184,
+      "grad_norm": 3.3032541275024414,
+      "learning_rate": 0.00011777777777777776,
+      "loss": 3.3843,
+      "step": 873
+    },
+    {
+      "epoch": 1.9389905712701054,
+      "grad_norm": 2.859706401824951,
+      "learning_rate": 0.00011753086419753086,
+      "loss": 3.393,
+      "step": 874
+    },
+    {
+      "epoch": 1.9412090959511925,
+      "grad_norm": 2.814079523086548,
+      "learning_rate": 0.00011728395061728394,
+      "loss": 3.6798,
+      "step": 875
+    },
+    {
+      "epoch": 1.9434276206322796,
+      "grad_norm": 2.7591936588287354,
+      "learning_rate": 0.00011703703703703702,
+      "loss": 3.2891,
+      "step": 876
+    },
+    {
+      "epoch": 1.9456461453133667,
+      "grad_norm": 2.8024144172668457,
+      "learning_rate": 0.0001167901234567901,
+      "loss": 3.3361,
+      "step": 877
+    },
+    {
+      "epoch": 1.9478646699944537,
+      "grad_norm": 2.9325807094573975,
+      "learning_rate": 0.00011654320987654319,
+      "loss": 3.2559,
+      "step": 878
+    },
+    {
+      "epoch": 1.9500831946755408,
+      "grad_norm": 2.514221429824829,
+      "learning_rate": 0.0001162962962962963,
+      "loss": 3.5964,
+      "step": 879
+    },
+    {
+      "epoch": 1.9523017193566279,
+      "grad_norm": 2.5519158840179443,
+      "learning_rate": 0.00011604938271604938,
+      "loss": 3.4779,
+      "step": 880
+    },
+    {
+      "epoch": 1.954520244037715,
+      "grad_norm": 2.764861583709717,
+      "learning_rate": 0.00011580246913580246,
+      "loss": 3.2935,
+      "step": 881
+    },
+    {
+      "epoch": 1.956738768718802,
+      "grad_norm": 2.798456907272339,
+      "learning_rate": 0.00011555555555555554,
+      "loss": 3.5363,
+      "step": 882
+    },
+    {
+      "epoch": 1.958957293399889,
+      "grad_norm": 2.389913558959961,
+      "learning_rate": 0.00011530864197530863,
+      "loss": 3.7013,
+      "step": 883
+    },
+    {
+      "epoch": 1.9611758180809762,
+      "grad_norm": 2.934868812561035,
+      "learning_rate": 0.00011506172839506172,
+      "loss": 3.3902,
+      "step": 884
+    },
+    {
+      "epoch": 1.9633943427620633,
+      "grad_norm": 3.020226001739502,
+      "learning_rate": 0.0001148148148148148,
+      "loss": 3.1595,
+      "step": 885
+    },
+    {
+      "epoch": 1.9656128674431503,
+      "grad_norm": 2.6467597484588623,
+      "learning_rate": 0.00011456790123456789,
+      "loss": 3.1879,
+      "step": 886
+    },
+    {
+      "epoch": 1.9678313921242374,
+      "grad_norm": 2.778237819671631,
+      "learning_rate": 0.00011432098765432099,
+      "loss": 3.804,
+      "step": 887
+    },
+    {
+      "epoch": 1.9700499168053245,
+      "grad_norm": 2.294633150100708,
+      "learning_rate": 0.00011407407407407407,
+      "loss": 3.4461,
+      "step": 888
+    },
+    {
+      "epoch": 1.9722684414864116,
+      "grad_norm": 2.6459619998931885,
+      "learning_rate": 0.00011382716049382715,
+      "loss": 3.5874,
+      "step": 889
+    },
+    {
+      "epoch": 1.9744869661674986,
+      "grad_norm": 2.783771276473999,
+      "learning_rate": 0.00011358024691358023,
+      "loss": 3.2295,
+      "step": 890
+    },
+    {
+      "epoch": 1.9767054908485857,
+      "grad_norm": 3.0905964374542236,
+      "learning_rate": 0.00011333333333333331,
+      "loss": 3.6168,
+      "step": 891
+    },
+    {
+      "epoch": 1.9789240155296728,
+      "grad_norm": 2.956162452697754,
+      "learning_rate": 0.00011308641975308641,
+      "loss": 3.4309,
+      "step": 892
+    },
+    {
+      "epoch": 1.9811425402107599,
+      "grad_norm": 2.762817859649658,
+      "learning_rate": 0.00011283950617283949,
+      "loss": 3.4602,
+      "step": 893
+    },
+    {
+      "epoch": 1.983361064891847,
+      "grad_norm": 2.8296544551849365,
+      "learning_rate": 0.00011259259259259259,
+      "loss": 3.5942,
+      "step": 894
+    },
+    {
+      "epoch": 1.985579589572934,
+      "grad_norm": 2.634073495864868,
+      "learning_rate": 0.00011234567901234567,
+      "loss": 3.6439,
+      "step": 895
+    },
+    {
+      "epoch": 1.987798114254021,
+      "grad_norm": 2.927875518798828,
+      "learning_rate": 0.00011209876543209876,
+      "loss": 3.2902,
+      "step": 896
+    },
+    {
+      "epoch": 1.9900166389351082,
+      "grad_norm": 2.3262245655059814,
+      "learning_rate": 0.00011185185185185184,
+      "loss": 3.0722,
+      "step": 897
+    },
+    {
+      "epoch": 1.9922351636161952,
+      "grad_norm": 2.763045310974121,
+      "learning_rate": 0.00011160493827160493,
+      "loss": 3.0764,
+      "step": 898
+    },
+    {
+      "epoch": 1.9944536882972823,
+      "grad_norm": 2.6812145709991455,
+      "learning_rate": 0.00011135802469135801,
+      "loss": 3.3975,
+      "step": 899
+    },
+    {
+      "epoch": 1.9966722129783694,
+      "grad_norm": 2.7727885246276855,
+      "learning_rate": 0.00011111111111111109,
+      "loss": 3.4061,
+      "step": 900
+    },
+    {
+      "epoch": 1.9988907376594565,
+      "grad_norm": 2.301645278930664,
+      "learning_rate": 0.0001108641975308642,
+      "loss": 3.4291,
+      "step": 901
+    },
+    {
+      "epoch": 2.0011092623405435,
+      "grad_norm": 2.214224338531494,
+      "learning_rate": 0.00011061728395061728,
+      "loss": 3.2183,
+      "step": 902
+    },
+    {
+      "epoch": 2.0033277870216306,
+      "grad_norm": 2.7368557453155518,
+      "learning_rate": 0.00011037037037037036,
+      "loss": 3.1214,
+      "step": 903
+    },
+    {
+      "epoch": 2.0055463117027177,
+      "grad_norm": 2.8103015422821045,
+      "learning_rate": 0.00011012345679012344,
+      "loss": 3.459,
+      "step": 904
+    },
+    {
+      "epoch": 2.0077648363838048,
+      "grad_norm": 2.6110711097717285,
+      "learning_rate": 0.00010987654320987654,
+      "loss": 3.522,
+      "step": 905
+    },
+    {
+      "epoch": 2.009983361064892,
+      "grad_norm": 3.3052570819854736,
+      "learning_rate": 0.00010962962962962962,
+      "loss": 3.3339,
+      "step": 906
+    },
+    {
+      "epoch": 2.012201885745979,
+      "grad_norm": 2.3670802116394043,
+      "learning_rate": 0.0001093827160493827,
+      "loss": 3.2402,
+      "step": 907
+    },
+    {
+      "epoch": 2.014420410427066,
+      "grad_norm": 2.4794681072235107,
+      "learning_rate": 0.00010913580246913578,
+      "loss": 3.405,
+      "step": 908
+    },
+    {
+      "epoch": 2.016638935108153,
+      "grad_norm": 2.895411968231201,
+      "learning_rate": 0.00010888888888888889,
+      "loss": 3.2727,
+      "step": 909
+    },
+    {
+      "epoch": 2.01885745978924,
+      "grad_norm": 3.0131473541259766,
+      "learning_rate": 0.00010864197530864197,
+      "loss": 3.3344,
+      "step": 910
+    },
+    {
+      "epoch": 2.021075984470327,
+      "grad_norm": 2.5086469650268555,
+      "learning_rate": 0.00010839506172839505,
+      "loss": 3.4074,
+      "step": 911
+    },
+    {
+      "epoch": 2.0232945091514143,
+      "grad_norm": 2.9425089359283447,
+      "learning_rate": 0.00010814814814814814,
+      "loss": 3.1704,
+      "step": 912
+    },
+    {
+      "epoch": 2.0255130338325014,
+      "grad_norm": 2.3922841548919678,
+      "learning_rate": 0.00010790123456790122,
+      "loss": 3.1185,
+      "step": 913
+    },
+    {
+      "epoch": 2.0277315585135884,
+      "grad_norm": 2.9616081714630127,
+      "learning_rate": 0.00010765432098765431,
+      "loss": 3.6164,
+      "step": 914
+    },
+    {
+      "epoch": 2.0299500831946755,
+      "grad_norm": 2.657465696334839,
+      "learning_rate": 0.0001074074074074074,
+      "loss": 3.2173,
+      "step": 915
+    },
+    {
+      "epoch": 2.0321686078757626,
+      "grad_norm": 3.143670082092285,
+      "learning_rate": 0.00010716049382716048,
+      "loss": 3.5582,
+      "step": 916
+    },
+    {
+      "epoch": 2.0343871325568497,
+      "grad_norm": 3.160522937774658,
+      "learning_rate": 0.00010691358024691357,
+      "loss": 3.2442,
+      "step": 917
+    },
+    {
+      "epoch": 2.0366056572379367,
+      "grad_norm": 2.589862823486328,
+      "learning_rate": 0.00010666666666666667,
+      "loss": 3.0435,
+      "step": 918
+    },
+    {
+      "epoch": 2.038824181919024,
+      "grad_norm": 2.1368284225463867,
+      "learning_rate": 0.00010641975308641975,
+      "loss": 3.3435,
+      "step": 919
+    },
+    {
+      "epoch": 2.041042706600111,
+      "grad_norm": 2.4189889430999756,
+      "learning_rate": 0.00010617283950617283,
+      "loss": 3.3012,
+      "step": 920
+    },
+    {
+      "epoch": 2.043261231281198,
+      "grad_norm": 3.2112157344818115,
+      "learning_rate": 0.00010592592592592591,
+      "loss": 3.2716,
+      "step": 921
+    },
+    {
+      "epoch": 2.045479755962285,
+      "grad_norm": 2.740027904510498,
+      "learning_rate": 0.00010567901234567899,
+      "loss": 3.4513,
+      "step": 922
+    },
+    {
+      "epoch": 2.047698280643372,
+      "grad_norm": 2.244269847869873,
+      "learning_rate": 0.00010543209876543209,
+      "loss": 3.3895,
+      "step": 923
+    },
+    {
+      "epoch": 2.049916805324459,
+      "grad_norm": 2.7902817726135254,
+      "learning_rate": 0.00010518518518518518,
+      "loss": 3.1981,
+      "step": 924
+    },
+    {
+      "epoch": 2.0521353300055463,
+      "grad_norm": 2.7596259117126465,
+      "learning_rate": 0.00010493827160493826,
+      "loss": 3.4054,
+      "step": 925
+    },
+    {
+      "epoch": 2.0543538546866333,
+      "grad_norm": 3.2806026935577393,
+      "learning_rate": 0.00010469135802469135,
+      "loss": 3.504,
+      "step": 926
+    },
+    {
+      "epoch": 2.0565723793677204,
+      "grad_norm": 2.1292428970336914,
+      "learning_rate": 0.00010444444444444444,
+      "loss": 3.3248,
+      "step": 927
+    },
+    {
+      "epoch": 2.0587909040488075,
+      "grad_norm": 2.7762598991394043,
+      "learning_rate": 0.00010419753086419752,
+      "loss": 3.1113,
+      "step": 928
+    },
+    {
+      "epoch": 2.0610094287298946,
+      "grad_norm": 3.3642425537109375,
+      "learning_rate": 0.0001039506172839506,
+      "loss": 3.3936,
+      "step": 929
+    },
+    {
+      "epoch": 2.0632279534109816,
+      "grad_norm": 2.5691473484039307,
+      "learning_rate": 0.00010370370370370369,
+      "loss": 3.3624,
+      "step": 930
+    },
+    {
+      "epoch": 2.0654464780920687,
+      "grad_norm": 2.7828006744384766,
+      "learning_rate": 0.00010345679012345677,
+      "loss": 3.4665,
+      "step": 931
+    },
+    {
+      "epoch": 2.067665002773156,
+      "grad_norm": 2.393995523452759,
+      "learning_rate": 0.00010320987654320988,
+      "loss": 3.142,
+      "step": 932
+    },
+    {
+      "epoch": 2.069883527454243,
+      "grad_norm": 2.6942689418792725,
+      "learning_rate": 0.00010296296296296296,
+      "loss": 3.2557,
+      "step": 933
+    },
+    {
+      "epoch": 2.07210205213533,
+      "grad_norm": 2.7326319217681885,
+      "learning_rate": 0.00010271604938271604,
+      "loss": 3.6079,
+      "step": 934
+    },
+    {
+      "epoch": 2.074320576816417,
+      "grad_norm": 2.867701768875122,
+      "learning_rate": 0.00010246913580246912,
+      "loss": 3.1579,
+      "step": 935
+    },
+    {
+      "epoch": 2.076539101497504,
+      "grad_norm": 2.5031862258911133,
+      "learning_rate": 0.00010222222222222222,
+      "loss": 3.4113,
+      "step": 936
+    },
+    {
+      "epoch": 2.078757626178591,
+      "grad_norm": 2.3299472332000732,
+      "learning_rate": 0.0001019753086419753,
+      "loss": 3.1299,
+      "step": 937
+    },
+    {
+      "epoch": 2.0809761508596782,
+      "grad_norm": 3.1566550731658936,
+      "learning_rate": 0.00010172839506172838,
+      "loss": 3.1598,
+      "step": 938
+    },
+    {
+      "epoch": 2.0831946755407653,
+      "grad_norm": 3.6371500492095947,
+      "learning_rate": 0.00010148148148148147,
+      "loss": 3.1183,
+      "step": 939
+    },
+    {
+      "epoch": 2.0854132002218524,
+      "grad_norm": 3.431079149246216,
+      "learning_rate": 0.00010123456790123457,
+      "loss": 3.3577,
+      "step": 940
+    },
+    {
+      "epoch": 2.0876317249029395,
+      "grad_norm": 2.6184544563293457,
+      "learning_rate": 0.00010098765432098765,
+      "loss": 3.3554,
+      "step": 941
+    },
+    {
+      "epoch": 2.0898502495840265,
+      "grad_norm": 2.5295379161834717,
+      "learning_rate": 0.00010074074074074073,
+      "loss": 3.2798,
+      "step": 942
+    },
+    {
+      "epoch": 2.0920687742651136,
+      "grad_norm": 2.5832324028015137,
+      "learning_rate": 0.00010049382716049381,
+      "loss": 3.0804,
+      "step": 943
+    },
+    {
+      "epoch": 2.0942872989462007,
+      "grad_norm": 2.876528024673462,
+      "learning_rate": 0.0001002469135802469,
+      "loss": 3.3908,
+      "step": 944
+    },
+    {
+      "epoch": 2.0965058236272878,
+      "grad_norm": 2.7915923595428467,
+      "learning_rate": 9.999999999999999e-05,
+      "loss": 3.5026,
+      "step": 945
+    },
+    {
+      "epoch": 2.098724348308375,
+      "grad_norm": 3.0117859840393066,
+      "learning_rate": 9.975308641975307e-05,
+      "loss": 3.3697,
+      "step": 946
+    },
+    {
+      "epoch": 2.100942872989462,
+      "grad_norm": 2.561058759689331,
+      "learning_rate": 9.950617283950617e-05,
+      "loss": 3.3587,
+      "step": 947
+    },
+    {
+      "epoch": 2.103161397670549,
+      "grad_norm": 2.8528170585632324,
+      "learning_rate": 9.925925925925925e-05,
+      "loss": 3.1795,
+      "step": 948
+    },
+    {
+      "epoch": 2.105379922351636,
+      "grad_norm": 2.910099983215332,
+      "learning_rate": 9.901234567901234e-05,
+      "loss": 3.4059,
+      "step": 949
+    },
+    {
+      "epoch": 2.107598447032723,
+      "grad_norm": 3.0954582691192627,
+      "learning_rate": 9.876543209876543e-05,
+      "loss": 3.2451,
+      "step": 950
+    },
+    {
+      "epoch": 2.10981697171381,
+      "grad_norm": 2.838718891143799,
+      "learning_rate": 9.851851851851851e-05,
+      "loss": 3.1884,
+      "step": 951
+    },
+    {
+      "epoch": 2.1120354963948973,
+      "grad_norm": 2.769526243209839,
+      "learning_rate": 9.827160493827159e-05,
+      "loss": 3.2314,
+      "step": 952
+    },
+    {
+      "epoch": 2.1142540210759844,
+      "grad_norm": 2.75699782371521,
+      "learning_rate": 9.802469135802467e-05,
+      "loss": 3.3221,
+      "step": 953
+    },
+    {
+      "epoch": 2.1164725457570714,
+      "grad_norm": 2.7146427631378174,
+      "learning_rate": 9.777777777777778e-05,
+      "loss": 3.4212,
+      "step": 954
+    },
+    {
+      "epoch": 2.1186910704381585,
+      "grad_norm": 2.674575090408325,
+      "learning_rate": 9.753086419753086e-05,
+      "loss": 3.3399,
+      "step": 955
+    },
+    {
+      "epoch": 2.1209095951192456,
+      "grad_norm": 2.636939287185669,
+      "learning_rate": 9.728395061728394e-05,
+      "loss": 3.5126,
+      "step": 956
+    },
+    {
+      "epoch": 2.1231281198003327,
+      "grad_norm": 2.4494354724884033,
+      "learning_rate": 9.703703703703702e-05,
+      "loss": 3.5247,
+      "step": 957
+    },
+    {
+      "epoch": 2.1253466444814197,
+      "grad_norm": 2.804647207260132,
+      "learning_rate": 9.679012345679012e-05,
+      "loss": 3.0891,
+      "step": 958
+    },
+    {
+      "epoch": 2.127565169162507,
+      "grad_norm": 2.5839295387268066,
+      "learning_rate": 9.65432098765432e-05,
+      "loss": 3.2709,
+      "step": 959
+    },
+    {
+      "epoch": 2.129783693843594,
+      "grad_norm": 2.7179274559020996,
+      "learning_rate": 9.629629629629628e-05,
+      "loss": 3.7235,
+      "step": 960
+    },
+    {
+      "epoch": 2.132002218524681,
+      "grad_norm": 2.8295676708221436,
+      "learning_rate": 9.604938271604936e-05,
+      "loss": 3.1686,
+      "step": 961
+    },
+    {
+      "epoch": 2.134220743205768,
+      "grad_norm": 3.2817068099975586,
+      "learning_rate": 9.580246913580247e-05,
+      "loss": 3.527,
+      "step": 962
+    },
+    {
+      "epoch": 2.136439267886855,
+      "grad_norm": 3.2478339672088623,
+      "learning_rate": 9.555555555555555e-05,
+      "loss": 3.5833,
+      "step": 963
+    },
+    {
+      "epoch": 2.138657792567942,
+      "grad_norm": 2.735914707183838,
+      "learning_rate": 9.530864197530864e-05,
+      "loss": 3.5107,
+      "step": 964
+    },
+    {
+      "epoch": 2.1408763172490293,
+      "grad_norm": 2.41105318069458,
+      "learning_rate": 9.506172839506172e-05,
+      "loss": 3.4533,
+      "step": 965
+    },
+    {
+      "epoch": 2.1430948419301163,
+      "grad_norm": 3.087629556655884,
+      "learning_rate": 9.48148148148148e-05,
+      "loss": 3.459,
+      "step": 966
+    },
+    {
+      "epoch": 2.1453133666112034,
+      "grad_norm": 2.7224557399749756,
+      "learning_rate": 9.45679012345679e-05,
+      "loss": 3.2979,
+      "step": 967
+    },
+    {
+      "epoch": 2.1475318912922905,
+      "grad_norm": 2.7184829711914062,
+      "learning_rate": 9.432098765432098e-05,
+      "loss": 3.6268,
+      "step": 968
+    },
+    {
+      "epoch": 2.1497504159733776,
+      "grad_norm": 2.501612663269043,
+      "learning_rate": 9.407407407407407e-05,
+      "loss": 3.4727,
+      "step": 969
+    },
+    {
+      "epoch": 2.1519689406544646,
+      "grad_norm": 3.000899314880371,
+      "learning_rate": 9.382716049382715e-05,
+      "loss": 3.2601,
+      "step": 970
+    },
+    {
+      "epoch": 2.1541874653355517,
+      "grad_norm": 4.00021505355835,
+      "learning_rate": 9.358024691358025e-05,
+      "loss": 3.1579,
+      "step": 971
+    },
+    {
+      "epoch": 2.156405990016639,
+      "grad_norm": 2.5587518215179443,
+      "learning_rate": 9.333333333333333e-05,
+      "loss": 3.0972,
+      "step": 972
+    },
+    {
+      "epoch": 2.158624514697726,
+      "grad_norm": 2.2487196922302246,
+      "learning_rate": 9.308641975308641e-05,
+      "loss": 3.3134,
+      "step": 973
+    },
+    {
+      "epoch": 2.160843039378813,
+      "grad_norm": 2.205644130706787,
+      "learning_rate": 9.283950617283949e-05,
+      "loss": 3.1762,
+      "step": 974
+    },
+    {
+      "epoch": 2.1630615640599,
+      "grad_norm": 2.336448907852173,
+      "learning_rate": 9.259259259259257e-05,
+      "loss": 3.551,
+      "step": 975
+    },
+    {
+      "epoch": 2.165280088740987,
+      "grad_norm": 2.489697217941284,
+      "learning_rate": 9.234567901234567e-05,
+      "loss": 3.2608,
+      "step": 976
+    },
+    {
+      "epoch": 2.167498613422074,
+      "grad_norm": 3.0866658687591553,
+      "learning_rate": 9.209876543209876e-05,
+      "loss": 3.3209,
+      "step": 977
+    },
+    {
+      "epoch": 2.1697171381031612,
+      "grad_norm": 2.650233268737793,
+      "learning_rate": 9.185185185185185e-05,
+      "loss": 3.1344,
+      "step": 978
+    },
+    {
+      "epoch": 2.1719356627842483,
+      "grad_norm": 2.72172474861145,
+      "learning_rate": 9.160493827160493e-05,
+      "loss": 3.5415,
+      "step": 979
+    },
+    {
+      "epoch": 2.1741541874653354,
+      "grad_norm": 2.992522716522217,
+      "learning_rate": 9.135802469135802e-05,
+      "loss": 3.3378,
+      "step": 980
+    },
+    {
+      "epoch": 2.1763727121464225,
+      "grad_norm": 2.6595747470855713,
+      "learning_rate": 9.11111111111111e-05,
+      "loss": 3.1837,
+      "step": 981
+    },
+    {
+      "epoch": 2.1785912368275095,
+      "grad_norm": 2.782557487487793,
+      "learning_rate": 9.086419753086419e-05,
+      "loss": 3.0579,
+      "step": 982
+    },
+    {
+      "epoch": 2.1808097615085966,
+      "grad_norm": 2.1831607818603516,
+      "learning_rate": 9.061728395061727e-05,
+      "loss": 3.7144,
+      "step": 983
+    },
+    {
+      "epoch": 2.1830282861896837,
+      "grad_norm": 3.1783084869384766,
+      "learning_rate": 9.037037037037038e-05,
+      "loss": 3.1371,
+      "step": 984
+    },
+    {
+      "epoch": 2.1852468108707708,
+      "grad_norm": 3.0358691215515137,
+      "learning_rate": 9.012345679012346e-05,
+      "loss": 3.5439,
+      "step": 985
+    },
+    {
+      "epoch": 2.187465335551858,
+      "grad_norm": 3.2464325428009033,
+      "learning_rate": 8.987654320987654e-05,
+      "loss": 3.5383,
+      "step": 986
+    },
+    {
+      "epoch": 2.189683860232945,
+      "grad_norm": 2.8475148677825928,
+      "learning_rate": 8.962962962962962e-05,
+      "loss": 2.8218,
+      "step": 987
+    },
+    {
+      "epoch": 2.191902384914032,
+      "grad_norm": 3.23492431640625,
+      "learning_rate": 8.93827160493827e-05,
+      "loss": 2.8587,
+      "step": 988
+    },
+    {
+      "epoch": 2.194120909595119,
+      "grad_norm": 2.9687252044677734,
+      "learning_rate": 8.91358024691358e-05,
+      "loss": 3.2085,
+      "step": 989
+    },
+    {
+      "epoch": 2.196339434276206,
+      "grad_norm": 2.781991481781006,
+      "learning_rate": 8.888888888888888e-05,
+      "loss": 3.0061,
+      "step": 990
+    },
+    {
+      "epoch": 2.1985579589572932,
+      "grad_norm": 3.5841755867004395,
+      "learning_rate": 8.864197530864196e-05,
+      "loss": 3.1554,
+      "step": 991
+    },
+    {
+      "epoch": 2.2007764836383803,
+      "grad_norm": 2.9188992977142334,
+      "learning_rate": 8.839506172839506e-05,
+      "loss": 3.4237,
+      "step": 992
+    },
+    {
+      "epoch": 2.2029950083194674,
+      "grad_norm": 3.3995234966278076,
+      "learning_rate": 8.814814814814815e-05,
+      "loss": 3.8444,
+      "step": 993
+    },
+    {
+      "epoch": 2.2052135330005544,
+      "grad_norm": 3.067474126815796,
+      "learning_rate": 8.790123456790123e-05,
+      "loss": 3.4538,
+      "step": 994
+    },
+    {
+      "epoch": 2.2074320576816415,
+      "grad_norm": 3.0776970386505127,
+      "learning_rate": 8.765432098765431e-05,
+      "loss": 3.2923,
+      "step": 995
+    },
+    {
+      "epoch": 2.2096505823627286,
+      "grad_norm": 2.702227830886841,
+      "learning_rate": 8.74074074074074e-05,
+      "loss": 3.1898,
+      "step": 996
+    },
+    {
+      "epoch": 2.2118691070438157,
+      "grad_norm": 2.4626901149749756,
+      "learning_rate": 8.716049382716048e-05,
+      "loss": 3.2111,
+      "step": 997
+    },
+    {
+      "epoch": 2.2140876317249027,
+      "grad_norm": 2.4106366634368896,
+      "learning_rate": 8.691358024691357e-05,
+      "loss": 3.4016,
+      "step": 998
+    },
+    {
+      "epoch": 2.21630615640599,
+      "grad_norm": 2.7973532676696777,
+      "learning_rate": 8.666666666666665e-05,
+      "loss": 2.8982,
+      "step": 999
+    },
+    {
+      "epoch": 2.218524681087077,
+      "grad_norm": 2.707000732421875,
+      "learning_rate": 8.641975308641975e-05,
+      "loss": 3.2077,
+      "step": 1000
+    },
+    {
+      "epoch": 2.220743205768164,
+      "grad_norm": 2.9762885570526123,
+      "learning_rate": 8.617283950617283e-05,
+      "loss": 3.1446,
+      "step": 1001
+    },
+    {
+      "epoch": 2.222961730449251,
+      "grad_norm": 2.851517915725708,
+      "learning_rate": 8.592592592592593e-05,
+      "loss": 3.4257,
+      "step": 1002
+    },
+    {
+      "epoch": 2.225180255130338,
+      "grad_norm": 3.141005516052246,
+      "learning_rate": 8.567901234567901e-05,
+      "loss": 3.1852,
+      "step": 1003
+    },
+    {
+      "epoch": 2.227398779811425,
+      "grad_norm": 2.6262354850769043,
+      "learning_rate": 8.543209876543209e-05,
+      "loss": 3.166,
+      "step": 1004
+    },
+    {
+      "epoch": 2.2296173044925123,
+      "grad_norm": 2.8922853469848633,
+      "learning_rate": 8.518518518518517e-05,
+      "loss": 3.1883,
+      "step": 1005
+    },
+    {
+      "epoch": 2.2318358291735994,
+      "grad_norm": 2.611370801925659,
+      "learning_rate": 8.493827160493825e-05,
+      "loss": 3.425,
+      "step": 1006
+    },
+    {
+      "epoch": 2.2340543538546864,
+      "grad_norm": 2.9298646450042725,
+      "learning_rate": 8.469135802469136e-05,
+      "loss": 3.4962,
+      "step": 1007
+    },
+    {
+      "epoch": 2.2362728785357735,
+      "grad_norm": 3.3863539695739746,
+      "learning_rate": 8.444444444444444e-05,
+      "loss": 3.5681,
+      "step": 1008
+    },
+    {
+      "epoch": 2.2384914032168606,
+      "grad_norm": 2.9136903285980225,
+      "learning_rate": 8.419753086419752e-05,
+      "loss": 3.6654,
+      "step": 1009
+    },
+    {
+      "epoch": 2.2407099278979477,
+      "grad_norm": 2.7800610065460205,
+      "learning_rate": 8.39506172839506e-05,
+      "loss": 3.452,
+      "step": 1010
+    },
+    {
+      "epoch": 2.2429284525790347,
+      "grad_norm": 3.5215325355529785,
+      "learning_rate": 8.37037037037037e-05,
+      "loss": 3.3121,
+      "step": 1011
+    },
+    {
+      "epoch": 2.245146977260122,
+      "grad_norm": 2.696180820465088,
+      "learning_rate": 8.345679012345678e-05,
+      "loss": 3.3153,
+      "step": 1012
+    },
+    {
+      "epoch": 2.247365501941209,
+      "grad_norm": 3.478760242462158,
+      "learning_rate": 8.320987654320986e-05,
+      "loss": 3.0322,
+      "step": 1013
+    },
+    {
+      "epoch": 2.249584026622296,
+      "grad_norm": 2.8756368160247803,
+      "learning_rate": 8.296296296296295e-05,
+      "loss": 3.6079,
+      "step": 1014
+    },
+    {
+      "epoch": 2.251802551303383,
+      "grad_norm": 3.254683494567871,
+      "learning_rate": 8.271604938271605e-05,
+      "loss": 3.6947,
+      "step": 1015
+    },
+    {
+      "epoch": 2.25402107598447,
+      "grad_norm": 2.7578141689300537,
+      "learning_rate": 8.246913580246914e-05,
+      "loss": 3.5089,
+      "step": 1016
+    },
+    {
+      "epoch": 2.256239600665557,
+      "grad_norm": 2.726099967956543,
+      "learning_rate": 8.222222222222222e-05,
+      "loss": 3.4293,
+      "step": 1017
+    },
+    {
+      "epoch": 2.2584581253466443,
+      "grad_norm": 2.8237040042877197,
+      "learning_rate": 8.19753086419753e-05,
+      "loss": 3.5336,
+      "step": 1018
+    },
+    {
+      "epoch": 2.2606766500277313,
+      "grad_norm": 2.735599994659424,
+      "learning_rate": 8.172839506172838e-05,
+      "loss": 3.3529,
+      "step": 1019
+    },
+    {
+      "epoch": 2.2628951747088184,
+      "grad_norm": 3.279878854751587,
+      "learning_rate": 8.148148148148148e-05,
+      "loss": 3.6011,
+      "step": 1020
+    },
+    {
+      "epoch": 2.2651136993899055,
+      "grad_norm": 3.0775787830352783,
+      "learning_rate": 8.123456790123456e-05,
+      "loss": 3.3948,
+      "step": 1021
+    },
+    {
+      "epoch": 2.2673322240709926,
+      "grad_norm": 2.7047524452209473,
+      "learning_rate": 8.098765432098765e-05,
+      "loss": 2.8768,
+      "step": 1022
+    },
+    {
+      "epoch": 2.2695507487520796,
+      "grad_norm": 2.916778802871704,
+      "learning_rate": 8.074074074074073e-05,
+      "loss": 3.3922,
+      "step": 1023
+    },
+    {
+      "epoch": 2.2717692734331667,
+      "grad_norm": 3.21169114112854,
+      "learning_rate": 8.049382716049383e-05,
+      "loss": 3.1699,
+      "step": 1024
+    },
+    {
+      "epoch": 2.2739877981142538,
+      "grad_norm": 2.75333571434021,
+      "learning_rate": 8.024691358024691e-05,
+      "loss": 3.4629,
+      "step": 1025
+    },
+    {
+      "epoch": 2.276206322795341,
+      "grad_norm": 3.6443674564361572,
+      "learning_rate": 7.999999999999999e-05,
+      "loss": 3.5277,
+      "step": 1026
+    },
+    {
+      "epoch": 2.278424847476428,
+      "grad_norm": 3.4318315982818604,
+      "learning_rate": 7.975308641975307e-05,
+      "loss": 3.0589,
+      "step": 1027
+    },
+    {
+      "epoch": 2.280643372157515,
+      "grad_norm": 3.112078905105591,
+      "learning_rate": 7.950617283950616e-05,
+      "loss": 3.3653,
+      "step": 1028
+    },
+    {
+      "epoch": 2.2828618968386025,
+      "grad_norm": 2.948298931121826,
+      "learning_rate": 7.925925925925925e-05,
+      "loss": 3.1545,
+      "step": 1029
+    },
+    {
+      "epoch": 2.2850804215196896,
+      "grad_norm": 2.7539749145507812,
+      "learning_rate": 7.901234567901235e-05,
+      "loss": 3.4846,
+      "step": 1030
+    },
+    {
+      "epoch": 2.2872989462007767,
+      "grad_norm": 2.9734842777252197,
+      "learning_rate": 7.876543209876543e-05,
+      "loss": 3.4928,
+      "step": 1031
+    },
+    {
+      "epoch": 2.2895174708818637,
+      "grad_norm": 3.1515142917633057,
+      "learning_rate": 7.851851851851851e-05,
+      "loss": 3.4928,
+      "step": 1032
+    },
+    {
+      "epoch": 2.291735995562951,
+      "grad_norm": 3.066748857498169,
+      "learning_rate": 7.82716049382716e-05,
+      "loss": 3.0412,
+      "step": 1033
+    },
+    {
+      "epoch": 2.293954520244038,
+      "grad_norm": 2.6987087726593018,
+      "learning_rate": 7.802469135802469e-05,
+      "loss": 3.3307,
+      "step": 1034
+    },
+    {
+      "epoch": 2.296173044925125,
+      "grad_norm": 3.103959798812866,
+      "learning_rate": 7.777777777777777e-05,
+      "loss": 3.4496,
+      "step": 1035
+    },
+    {
+      "epoch": 2.298391569606212,
+      "grad_norm": 2.627352237701416,
+      "learning_rate": 7.753086419753085e-05,
+      "loss": 3.5156,
+      "step": 1036
+    },
+    {
+      "epoch": 2.300610094287299,
+      "grad_norm": 3.050983428955078,
+      "learning_rate": 7.728395061728396e-05,
+      "loss": 3.0821,
+      "step": 1037
+    },
+    {
+      "epoch": 2.302828618968386,
+      "grad_norm": 3.339452028274536,
+      "learning_rate": 7.703703703703704e-05,
+      "loss": 3.5296,
+      "step": 1038
+    },
+    {
+      "epoch": 2.3050471436494733,
+      "grad_norm": 3.8552327156066895,
+      "learning_rate": 7.679012345679012e-05,
+      "loss": 3.0475,
+      "step": 1039
+    },
+    {
+      "epoch": 2.3072656683305603,
+      "grad_norm": 3.0647966861724854,
+      "learning_rate": 7.65432098765432e-05,
+      "loss": 3.231,
+      "step": 1040
+    },
+    {
+      "epoch": 2.3094841930116474,
+      "grad_norm": 2.5461511611938477,
+      "learning_rate": 7.629629629629628e-05,
+      "loss": 3.5408,
+      "step": 1041
+    },
+    {
+      "epoch": 2.3117027176927345,
+      "grad_norm": 2.9686331748962402,
+      "learning_rate": 7.604938271604937e-05,
+      "loss": 3.4201,
+      "step": 1042
+    },
+    {
+      "epoch": 2.3139212423738216,
+      "grad_norm": 2.692967653274536,
+      "learning_rate": 7.580246913580246e-05,
+      "loss": 3.1088,
+      "step": 1043
+    },
+    {
+      "epoch": 2.3161397670549086,
+      "grad_norm": 2.325434684753418,
+      "learning_rate": 7.555555555555554e-05,
+      "loss": 3.0657,
+      "step": 1044
+    },
+    {
+      "epoch": 2.3183582917359957,
+      "grad_norm": 2.974717378616333,
+      "learning_rate": 7.530864197530864e-05,
+      "loss": 3.4904,
+      "step": 1045
+    },
+    {
+      "epoch": 2.320576816417083,
+      "grad_norm": 2.145707845687866,
+      "learning_rate": 7.506172839506173e-05,
+      "loss": 3.3979,
+      "step": 1046
+    },
+    {
+      "epoch": 2.32279534109817,
+      "grad_norm": 2.3607261180877686,
+      "learning_rate": 7.481481481481481e-05,
+      "loss": 3.2763,
+      "step": 1047
+    },
+    {
+      "epoch": 2.325013865779257,
+      "grad_norm": 2.9835612773895264,
+      "learning_rate": 7.45679012345679e-05,
+      "loss": 3.154,
+      "step": 1048
+    },
+    {
+      "epoch": 2.327232390460344,
+      "grad_norm": 3.2200374603271484,
+      "learning_rate": 7.432098765432098e-05,
+      "loss": 3.0069,
+      "step": 1049
+    },
+    {
+      "epoch": 2.329450915141431,
+      "grad_norm": 2.755783796310425,
+      "learning_rate": 7.407407407407406e-05,
+      "loss": 3.182,
+      "step": 1050
+    },
+    {
+      "epoch": 2.331669439822518,
+      "grad_norm": 2.6484782695770264,
+      "learning_rate": 7.382716049382715e-05,
+      "loss": 3.2957,
+      "step": 1051
+    },
+    {
+      "epoch": 2.3338879645036053,
+      "grad_norm": 2.7825443744659424,
+      "learning_rate": 7.358024691358024e-05,
+      "loss": 3.0865,
+      "step": 1052
+    },
+    {
+      "epoch": 2.3361064891846923,
+      "grad_norm": 3.4233875274658203,
+      "learning_rate": 7.333333333333332e-05,
+      "loss": 3.5636,
+      "step": 1053
+    },
+    {
+      "epoch": 2.3383250138657794,
+      "grad_norm": 2.289304494857788,
+      "learning_rate": 7.308641975308641e-05,
+      "loss": 3.1869,
+      "step": 1054
+    },
+    {
+      "epoch": 2.3405435385468665,
+      "grad_norm": 2.7351839542388916,
+      "learning_rate": 7.28395061728395e-05,
+      "loss": 3.1327,
+      "step": 1055
+    },
+    {
+      "epoch": 2.3427620632279536,
+      "grad_norm": 3.2106435298919678,
+      "learning_rate": 7.259259259259259e-05,
+      "loss": 3.5141,
+      "step": 1056
+    },
+    {
+      "epoch": 2.3449805879090406,
+      "grad_norm": 3.0602219104766846,
+      "learning_rate": 7.234567901234567e-05,
+      "loss": 3.2521,
+      "step": 1057
+    },
+    {
+      "epoch": 2.3471991125901277,
+      "grad_norm": 2.7595155239105225,
+      "learning_rate": 7.209876543209877e-05,
+      "loss": 3.441,
+      "step": 1058
+    },
+    {
+      "epoch": 2.3494176372712148,
+      "grad_norm": 2.665210008621216,
+      "learning_rate": 7.185185185185185e-05,
+      "loss": 3.2033,
+      "step": 1059
+    },
+    {
+      "epoch": 2.351636161952302,
+      "grad_norm": 3.1021833419799805,
+      "learning_rate": 7.160493827160493e-05,
+      "loss": 3.0613,
+      "step": 1060
+    },
+    {
+      "epoch": 2.353854686633389,
+      "grad_norm": 2.6876745223999023,
+      "learning_rate": 7.135802469135801e-05,
+      "loss": 3.2604,
+      "step": 1061
+    },
+    {
+      "epoch": 2.356073211314476,
+      "grad_norm": 2.7640628814697266,
+      "learning_rate": 7.11111111111111e-05,
+      "loss": 3.1752,
+      "step": 1062
+    },
+    {
+      "epoch": 2.358291735995563,
+      "grad_norm": 2.6364986896514893,
+      "learning_rate": 7.086419753086419e-05,
+      "loss": 3.2047,
+      "step": 1063
+    },
+    {
+      "epoch": 2.36051026067665,
+      "grad_norm": 2.878695011138916,
+      "learning_rate": 7.061728395061727e-05,
+      "loss": 3.2983,
+      "step": 1064
+    },
+    {
+      "epoch": 2.3627287853577372,
+      "grad_norm": 3.2854676246643066,
+      "learning_rate": 7.037037037037036e-05,
+      "loss": 3.1613,
+      "step": 1065
+    },
+    {
+      "epoch": 2.3649473100388243,
+      "grad_norm": 2.446117877960205,
+      "learning_rate": 7.012345679012345e-05,
+      "loss": 3.0687,
+      "step": 1066
+    },
+    {
+      "epoch": 2.3671658347199114,
+      "grad_norm": 2.967456817626953,
+      "learning_rate": 6.987654320987654e-05,
+      "loss": 3.0897,
+      "step": 1067
+    },
+    {
+      "epoch": 2.3693843594009985,
+      "grad_norm": 2.4819982051849365,
+      "learning_rate": 6.962962962962962e-05,
+      "loss": 3.353,
+      "step": 1068
+    },
+    {
+      "epoch": 2.3716028840820855,
+      "grad_norm": 3.0545310974121094,
+      "learning_rate": 6.938271604938272e-05,
+      "loss": 3.4094,
+      "step": 1069
+    },
+    {
+      "epoch": 2.3738214087631726,
+      "grad_norm": 3.144083261489868,
+      "learning_rate": 6.91358024691358e-05,
+      "loss": 3.192,
+      "step": 1070
+    },
+    {
+      "epoch": 2.3760399334442597,
+      "grad_norm": 2.700172185897827,
+      "learning_rate": 6.888888888888888e-05,
+      "loss": 3.5315,
+      "step": 1071
+    },
+    {
+      "epoch": 2.3782584581253468,
+      "grad_norm": 3.2784016132354736,
+      "learning_rate": 6.864197530864196e-05,
+      "loss": 3.3584,
+      "step": 1072
+    },
+    {
+      "epoch": 2.380476982806434,
+      "grad_norm": 2.5297181606292725,
+      "learning_rate": 6.839506172839506e-05,
+      "loss": 3.477,
+      "step": 1073
+    },
+    {
+      "epoch": 2.382695507487521,
+      "grad_norm": 2.6563522815704346,
+      "learning_rate": 6.814814814814814e-05,
+      "loss": 3.2554,
+      "step": 1074
+    },
+    {
+      "epoch": 2.384914032168608,
+      "grad_norm": 4.103996276855469,
+      "learning_rate": 6.790123456790122e-05,
+      "loss": 2.953,
+      "step": 1075
+    },
+    {
+      "epoch": 2.387132556849695,
+      "grad_norm": 2.9415171146392822,
+      "learning_rate": 6.765432098765432e-05,
+      "loss": 3.3589,
+      "step": 1076
+    },
+    {
+      "epoch": 2.389351081530782,
+      "grad_norm": 2.971709728240967,
+      "learning_rate": 6.74074074074074e-05,
+      "loss": 3.8504,
+      "step": 1077
+    },
+    {
+      "epoch": 2.391569606211869,
+      "grad_norm": 2.378976821899414,
+      "learning_rate": 6.716049382716049e-05,
+      "loss": 3.5055,
+      "step": 1078
+    },
+    {
+      "epoch": 2.3937881308929563,
+      "grad_norm": 3.322303056716919,
+      "learning_rate": 6.691358024691357e-05,
+      "loss": 3.2525,
+      "step": 1079
+    },
+    {
+      "epoch": 2.3960066555740434,
+      "grad_norm": 2.9601850509643555,
+      "learning_rate": 6.666666666666666e-05,
+      "loss": 3.4699,
+      "step": 1080
+    },
+    {
+      "epoch": 2.3982251802551304,
+      "grad_norm": 2.8059635162353516,
+      "learning_rate": 6.641975308641975e-05,
+      "loss": 3.264,
+      "step": 1081
+    },
+    {
+      "epoch": 2.4004437049362175,
+      "grad_norm": 2.5180535316467285,
+      "learning_rate": 6.617283950617283e-05,
+      "loss": 3.2555,
+      "step": 1082
+    },
+    {
+      "epoch": 2.4026622296173046,
+      "grad_norm": 2.2536442279815674,
+      "learning_rate": 6.592592592592591e-05,
+      "loss": 3.3812,
+      "step": 1083
+    },
+    {
+      "epoch": 2.4048807542983917,
+      "grad_norm": 2.898268938064575,
+      "learning_rate": 6.567901234567901e-05,
+      "loss": 3.252,
+      "step": 1084
+    },
+    {
+      "epoch": 2.4070992789794787,
+      "grad_norm": 2.6837446689605713,
+      "learning_rate": 6.543209876543209e-05,
+      "loss": 3.3256,
+      "step": 1085
+    },
+    {
+      "epoch": 2.409317803660566,
+      "grad_norm": 2.5627522468566895,
+      "learning_rate": 6.518518518518517e-05,
+      "loss": 3.0431,
+      "step": 1086
+    },
+    {
+      "epoch": 2.411536328341653,
+      "grad_norm": 2.8901147842407227,
+      "learning_rate": 6.493827160493827e-05,
+      "loss": 3.1594,
+      "step": 1087
+    },
+    {
+      "epoch": 2.41375485302274,
+      "grad_norm": 3.0189077854156494,
+      "learning_rate": 6.469135802469135e-05,
+      "loss": 3.117,
+      "step": 1088
+    },
+    {
+      "epoch": 2.415973377703827,
+      "grad_norm": 2.7954907417297363,
+      "learning_rate": 6.444444444444444e-05,
+      "loss": 3.2294,
+      "step": 1089
+    },
+    {
+      "epoch": 2.418191902384914,
+      "grad_norm": 2.8154962062835693,
+      "learning_rate": 6.419753086419753e-05,
+      "loss": 3.3108,
+      "step": 1090
+    },
+    {
+      "epoch": 2.420410427066001,
+      "grad_norm": 3.1895527839660645,
+      "learning_rate": 6.395061728395061e-05,
+      "loss": 3.3627,
+      "step": 1091
+    },
+    {
+      "epoch": 2.4226289517470883,
+      "grad_norm": 2.3513667583465576,
+      "learning_rate": 6.37037037037037e-05,
+      "loss": 3.2149,
+      "step": 1092
+    },
+    {
+      "epoch": 2.4248474764281753,
+      "grad_norm": 2.9598026275634766,
+      "learning_rate": 6.345679012345678e-05,
+      "loss": 3.3555,
+      "step": 1093
+    },
+    {
+      "epoch": 2.4270660011092624,
+      "grad_norm": 3.145817756652832,
+      "learning_rate": 6.320987654320987e-05,
+      "loss": 3.4581,
+      "step": 1094
+    },
+    {
+      "epoch": 2.4292845257903495,
+      "grad_norm": 2.596997022628784,
+      "learning_rate": 6.296296296296295e-05,
+      "loss": 3.4155,
+      "step": 1095
+    },
+    {
+      "epoch": 2.4315030504714366,
+      "grad_norm": 2.6237974166870117,
+      "learning_rate": 6.271604938271604e-05,
+      "loss": 3.3958,
+      "step": 1096
+    },
+    {
+      "epoch": 2.4337215751525236,
+      "grad_norm": 2.7650225162506104,
+      "learning_rate": 6.246913580246912e-05,
+      "loss": 3.1428,
+      "step": 1097
+    },
+    {
+      "epoch": 2.4359400998336107,
+      "grad_norm": 3.4287047386169434,
+      "learning_rate": 6.222222222222222e-05,
+      "loss": 3.2351,
+      "step": 1098
+    },
+    {
+      "epoch": 2.438158624514698,
+      "grad_norm": 2.8054420948028564,
+      "learning_rate": 6.19753086419753e-05,
+      "loss": 3.1093,
+      "step": 1099
+    },
+    {
+      "epoch": 2.440377149195785,
+      "grad_norm": 2.7483785152435303,
+      "learning_rate": 6.17283950617284e-05,
+      "loss": 3.2448,
+      "step": 1100
+    },
+    {
+      "epoch": 2.442595673876872,
+      "grad_norm": 2.506194829940796,
+      "learning_rate": 6.148148148148148e-05,
+      "loss": 3.285,
+      "step": 1101
+    },
+    {
+      "epoch": 2.444814198557959,
+      "grad_norm": 3.0636210441589355,
+      "learning_rate": 6.123456790123456e-05,
+      "loss": 3.1988,
+      "step": 1102
+    },
+    {
+      "epoch": 2.447032723239046,
+      "grad_norm": 2.5065407752990723,
+      "learning_rate": 6.0987654320987655e-05,
+      "loss": 3.1644,
+      "step": 1103
+    },
+    {
+      "epoch": 2.449251247920133,
+      "grad_norm": 2.4532384872436523,
+      "learning_rate": 6.0740740740740736e-05,
+      "loss": 3.1665,
+      "step": 1104
+    },
+    {
+      "epoch": 2.4514697726012202,
+      "grad_norm": 3.2283921241760254,
+      "learning_rate": 6.049382716049382e-05,
+      "loss": 3.1345,
+      "step": 1105
+    },
+    {
+      "epoch": 2.4536882972823073,
+      "grad_norm": 2.8355252742767334,
+      "learning_rate": 6.0246913580246906e-05,
+      "loss": 3.0779,
+      "step": 1106
+    },
+    {
+      "epoch": 2.4559068219633944,
+      "grad_norm": 2.997873544692993,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 3.4586,
+      "step": 1107
+    },
+    {
+      "epoch": 2.4581253466444815,
+      "grad_norm": 2.386368989944458,
+      "learning_rate": 5.975308641975308e-05,
+      "loss": 3.044,
+      "step": 1108
+    },
+    {
+      "epoch": 2.4603438713255685,
+      "grad_norm": 2.8367013931274414,
+      "learning_rate": 5.9506172839506165e-05,
+      "loss": 2.9237,
+      "step": 1109
+    },
+    {
+      "epoch": 2.4625623960066556,
+      "grad_norm": 3.036980390548706,
+      "learning_rate": 5.925925925925925e-05,
+      "loss": 3.3497,
+      "step": 1110
+    },
+    {
+      "epoch": 2.4647809206877427,
+      "grad_norm": 2.8131797313690186,
+      "learning_rate": 5.901234567901234e-05,
+      "loss": 3.1857,
+      "step": 1111
+    },
+    {
+      "epoch": 2.4669994453688298,
+      "grad_norm": 2.419877052307129,
+      "learning_rate": 5.876543209876543e-05,
+      "loss": 3.0645,
+      "step": 1112
+    },
+    {
+      "epoch": 2.469217970049917,
+      "grad_norm": 2.9670376777648926,
+      "learning_rate": 5.851851851851851e-05,
+      "loss": 3.329,
+      "step": 1113
+    },
+    {
+      "epoch": 2.471436494731004,
+      "grad_norm": 2.5262067317962646,
+      "learning_rate": 5.827160493827159e-05,
+      "loss": 3.409,
+      "step": 1114
+    },
+    {
+      "epoch": 2.473655019412091,
+      "grad_norm": 2.950669050216675,
+      "learning_rate": 5.802469135802469e-05,
+      "loss": 2.9612,
+      "step": 1115
+    },
+    {
+      "epoch": 2.475873544093178,
+      "grad_norm": 3.055124521255493,
+      "learning_rate": 5.777777777777777e-05,
+      "loss": 3.519,
+      "step": 1116
+    },
+    {
+      "epoch": 2.478092068774265,
+      "grad_norm": 2.6472926139831543,
+      "learning_rate": 5.753086419753086e-05,
+      "loss": 3.2575,
+      "step": 1117
+    },
+    {
+      "epoch": 2.480310593455352,
+      "grad_norm": 2.971778392791748,
+      "learning_rate": 5.7283950617283946e-05,
+      "loss": 3.1498,
+      "step": 1118
+    },
+    {
+      "epoch": 2.4825291181364393,
+      "grad_norm": 3.0063676834106445,
+      "learning_rate": 5.7037037037037035e-05,
+      "loss": 3.3592,
+      "step": 1119
+    },
+    {
+      "epoch": 2.4847476428175264,
+      "grad_norm": 2.7526888847351074,
+      "learning_rate": 5.6790123456790116e-05,
+      "loss": 3.353,
+      "step": 1120
+    },
+    {
+      "epoch": 2.4869661674986134,
+      "grad_norm": 3.113614320755005,
+      "learning_rate": 5.6543209876543205e-05,
+      "loss": 3.3614,
+      "step": 1121
+    },
+    {
+      "epoch": 2.4891846921797005,
+      "grad_norm": 2.7114267349243164,
+      "learning_rate": 5.629629629629629e-05,
+      "loss": 3.2148,
+      "step": 1122
+    },
+    {
+      "epoch": 2.4914032168607876,
+      "grad_norm": 3.167954921722412,
+      "learning_rate": 5.604938271604938e-05,
+      "loss": 3.0925,
+      "step": 1123
+    },
+    {
+      "epoch": 2.4936217415418747,
+      "grad_norm": 3.2006332874298096,
+      "learning_rate": 5.580246913580246e-05,
+      "loss": 3.2694,
+      "step": 1124
+    },
+    {
+      "epoch": 2.4958402662229617,
+      "grad_norm": 2.7918267250061035,
+      "learning_rate": 5.5555555555555545e-05,
+      "loss": 3.7266,
+      "step": 1125
+    },
+    {
+      "epoch": 2.498058790904049,
+      "grad_norm": 2.5595762729644775,
+      "learning_rate": 5.530864197530864e-05,
+      "loss": 3.5575,
+      "step": 1126
+    },
+    {
+      "epoch": 2.500277315585136,
+      "grad_norm": 2.7684407234191895,
+      "learning_rate": 5.506172839506172e-05,
+      "loss": 3.1343,
+      "step": 1127
+    },
+    {
+      "epoch": 2.502495840266223,
+      "grad_norm": 2.877819061279297,
+      "learning_rate": 5.481481481481481e-05,
+      "loss": 3.1394,
+      "step": 1128
+    },
+    {
+      "epoch": 2.50471436494731,
+      "grad_norm": 2.793877124786377,
+      "learning_rate": 5.456790123456789e-05,
+      "loss": 3.434,
+      "step": 1129
+    },
+    {
+      "epoch": 2.506932889628397,
+      "grad_norm": 3.288165330886841,
+      "learning_rate": 5.4320987654320986e-05,
+      "loss": 3.1942,
+      "step": 1130
+    },
+    {
+      "epoch": 2.509151414309484,
+      "grad_norm": 2.9898035526275635,
+      "learning_rate": 5.407407407407407e-05,
+      "loss": 3.4179,
+      "step": 1131
+    },
+    {
+      "epoch": 2.5113699389905713,
+      "grad_norm": 2.4935641288757324,
+      "learning_rate": 5.3827160493827156e-05,
+      "loss": 3.2072,
+      "step": 1132
+    },
+    {
+      "epoch": 2.5135884636716583,
+      "grad_norm": 2.666517496109009,
+      "learning_rate": 5.358024691358024e-05,
+      "loss": 3.2782,
+      "step": 1133
+    },
+    {
+      "epoch": 2.5158069883527454,
+      "grad_norm": 2.5118372440338135,
+      "learning_rate": 5.333333333333333e-05,
+      "loss": 3.4282,
+      "step": 1134
+    },
+    {
+      "epoch": 2.5180255130338325,
+      "grad_norm": 2.647932529449463,
+      "learning_rate": 5.3086419753086415e-05,
+      "loss": 3.1268,
+      "step": 1135
+    },
+    {
+      "epoch": 2.5202440377149196,
+      "grad_norm": 2.876624584197998,
+      "learning_rate": 5.2839506172839496e-05,
+      "loss": 3.3554,
+      "step": 1136
+    },
+    {
+      "epoch": 2.5224625623960066,
+      "grad_norm": 3.057648181915283,
+      "learning_rate": 5.259259259259259e-05,
+      "loss": 3.3657,
+      "step": 1137
+    },
+    {
+      "epoch": 2.5246810870770937,
+      "grad_norm": 2.830214262008667,
+      "learning_rate": 5.234567901234567e-05,
+      "loss": 3.3393,
+      "step": 1138
+    },
+    {
+      "epoch": 2.526899611758181,
+      "grad_norm": 2.7469160556793213,
+      "learning_rate": 5.209876543209876e-05,
+      "loss": 3.5493,
+      "step": 1139
+    },
+    {
+      "epoch": 2.529118136439268,
+      "grad_norm": 2.8592453002929688,
+      "learning_rate": 5.185185185185184e-05,
+      "loss": 3.2821,
+      "step": 1140
+    },
+    {
+      "epoch": 2.531336661120355,
+      "grad_norm": 3.789799213409424,
+      "learning_rate": 5.160493827160494e-05,
+      "loss": 3.0911,
+      "step": 1141
+    },
+    {
+      "epoch": 2.533555185801442,
+      "grad_norm": 2.9866318702697754,
+      "learning_rate": 5.135802469135802e-05,
+      "loss": 3.1961,
+      "step": 1142
+    },
+    {
+      "epoch": 2.535773710482529,
+      "grad_norm": 2.3589515686035156,
+      "learning_rate": 5.111111111111111e-05,
+      "loss": 3.3896,
+      "step": 1143
+    },
+    {
+      "epoch": 2.537992235163616,
+      "grad_norm": 2.4686408042907715,
+      "learning_rate": 5.086419753086419e-05,
+      "loss": 3.5895,
+      "step": 1144
+    },
+    {
+      "epoch": 2.5402107598447032,
+      "grad_norm": 2.5708396434783936,
+      "learning_rate": 5.0617283950617285e-05,
+      "loss": 3.1888,
+      "step": 1145
+    },
+    {
+      "epoch": 2.5424292845257903,
+      "grad_norm": 2.8607640266418457,
+      "learning_rate": 5.0370370370370366e-05,
+      "loss": 3.3075,
+      "step": 1146
+    },
+    {
+      "epoch": 2.5446478092068774,
+      "grad_norm": 2.690152883529663,
+      "learning_rate": 5.012345679012345e-05,
+      "loss": 3.4681,
+      "step": 1147
+    },
+    {
+      "epoch": 2.5468663338879645,
+      "grad_norm": 2.6835904121398926,
+      "learning_rate": 4.9876543209876536e-05,
+      "loss": 3.3379,
+      "step": 1148
+    },
+    {
+      "epoch": 2.5490848585690515,
+      "grad_norm": 2.871027708053589,
+      "learning_rate": 4.9629629629629625e-05,
+      "loss": 3.3502,
+      "step": 1149
+    },
+    {
+      "epoch": 2.5513033832501386,
+      "grad_norm": 2.940917730331421,
+      "learning_rate": 4.938271604938271e-05,
+      "loss": 3.1446,
+      "step": 1150
+    },
+    {
+      "epoch": 2.5535219079312257,
+      "grad_norm": 2.5946738719940186,
+      "learning_rate": 4.9135802469135795e-05,
+      "loss": 3.1492,
+      "step": 1151
+    },
+    {
+      "epoch": 2.5557404326123128,
+      "grad_norm": 3.0175821781158447,
+      "learning_rate": 4.888888888888889e-05,
+      "loss": 3.1005,
+      "step": 1152
+    },
+    {
+      "epoch": 2.5579589572934,
+      "grad_norm": 3.2317311763763428,
+      "learning_rate": 4.864197530864197e-05,
+      "loss": 3.3764,
+      "step": 1153
+    },
+    {
+      "epoch": 2.560177481974487,
+      "grad_norm": 2.9672257900238037,
+      "learning_rate": 4.839506172839506e-05,
+      "loss": 3.5418,
+      "step": 1154
+    },
+    {
+      "epoch": 2.562396006655574,
+      "grad_norm": 2.80869722366333,
+      "learning_rate": 4.814814814814814e-05,
+      "loss": 3.336,
+      "step": 1155
+    },
+    {
+      "epoch": 2.564614531336661,
+      "grad_norm": 2.858304738998413,
+      "learning_rate": 4.7901234567901237e-05,
+      "loss": 3.1383,
+      "step": 1156
+    },
+    {
+      "epoch": 2.566833056017748,
+      "grad_norm": 2.749044418334961,
+      "learning_rate": 4.765432098765432e-05,
+      "loss": 3.1139,
+      "step": 1157
+    },
+    {
+      "epoch": 2.569051580698835,
+      "grad_norm": 2.8280067443847656,
+      "learning_rate": 4.74074074074074e-05,
+      "loss": 3.5982,
+      "step": 1158
+    },
+    {
+      "epoch": 2.5712701053799223,
+      "grad_norm": 2.9874823093414307,
+      "learning_rate": 4.716049382716049e-05,
+      "loss": 3.1597,
+      "step": 1159
+    },
+    {
+      "epoch": 2.5734886300610094,
+      "grad_norm": 3.2048020362854004,
+      "learning_rate": 4.6913580246913576e-05,
+      "loss": 3.1973,
+      "step": 1160
+    },
+    {
+      "epoch": 2.5757071547420964,
+      "grad_norm": 3.1794228553771973,
+      "learning_rate": 4.6666666666666665e-05,
+      "loss": 3.5809,
+      "step": 1161
+    },
+    {
+      "epoch": 2.5779256794231835,
+      "grad_norm": 2.7882235050201416,
+      "learning_rate": 4.6419753086419746e-05,
+      "loss": 3.2233,
+      "step": 1162
+    },
+    {
+      "epoch": 2.5801442041042706,
+      "grad_norm": 2.6559722423553467,
+      "learning_rate": 4.6172839506172835e-05,
+      "loss": 3.1751,
+      "step": 1163
+    },
+    {
+      "epoch": 2.5823627287853577,
+      "grad_norm": 2.8107988834381104,
+      "learning_rate": 4.592592592592592e-05,
+      "loss": 3.323,
+      "step": 1164
+    },
+    {
+      "epoch": 2.5845812534664447,
+      "grad_norm": 2.5513288974761963,
+      "learning_rate": 4.567901234567901e-05,
+      "loss": 3.232,
+      "step": 1165
+    },
+    {
+      "epoch": 2.586799778147532,
+      "grad_norm": 3.045684814453125,
+      "learning_rate": 4.543209876543209e-05,
+      "loss": 3.4789,
+      "step": 1166
+    },
+    {
+      "epoch": 2.589018302828619,
+      "grad_norm": 2.9420974254608154,
+      "learning_rate": 4.518518518518519e-05,
+      "loss": 3.4336,
+      "step": 1167
+    },
+    {
+      "epoch": 2.591236827509706,
+      "grad_norm": 2.5203821659088135,
+      "learning_rate": 4.493827160493827e-05,
+      "loss": 3.0736,
+      "step": 1168
+    },
+    {
+      "epoch": 2.593455352190793,
+      "grad_norm": 2.601893901824951,
+      "learning_rate": 4.469135802469135e-05,
+      "loss": 3.0939,
+      "step": 1169
+    },
+    {
+      "epoch": 2.59567387687188,
+      "grad_norm": 2.7631258964538574,
+      "learning_rate": 4.444444444444444e-05,
+      "loss": 3.3785,
+      "step": 1170
+    },
+    {
+      "epoch": 2.597892401552967,
+      "grad_norm": 2.8595950603485107,
+      "learning_rate": 4.419753086419753e-05,
+      "loss": 3.0413,
+      "step": 1171
+    },
+    {
+      "epoch": 2.6001109262340543,
+      "grad_norm": 2.522836208343506,
+      "learning_rate": 4.3950617283950617e-05,
+      "loss": 3.5467,
+      "step": 1172
+    },
+    {
+      "epoch": 2.6023294509151413,
+      "grad_norm": 2.840735912322998,
+      "learning_rate": 4.37037037037037e-05,
+      "loss": 3.0608,
+      "step": 1173
+    },
+    {
+      "epoch": 2.6045479755962284,
+      "grad_norm": 2.7334539890289307,
+      "learning_rate": 4.3456790123456786e-05,
+      "loss": 3.1843,
+      "step": 1174
+    },
+    {
+      "epoch": 2.6067665002773155,
+      "grad_norm": 3.136683702468872,
+      "learning_rate": 4.3209876543209875e-05,
+      "loss": 3.2086,
+      "step": 1175
+    },
+    {
+      "epoch": 2.6089850249584026,
+      "grad_norm": 2.7305057048797607,
+      "learning_rate": 4.296296296296296e-05,
+      "loss": 3.224,
+      "step": 1176
+    },
+    {
+      "epoch": 2.6112035496394896,
+      "grad_norm": 2.394388437271118,
+      "learning_rate": 4.2716049382716045e-05,
+      "loss": 3.3941,
+      "step": 1177
+    },
+    {
+      "epoch": 2.6134220743205767,
+      "grad_norm": 2.4909205436706543,
+      "learning_rate": 4.2469135802469126e-05,
+      "loss": 3.137,
+      "step": 1178
+    },
+    {
+      "epoch": 2.615640599001664,
+      "grad_norm": 2.694131374359131,
+      "learning_rate": 4.222222222222222e-05,
+      "loss": 3.4478,
+      "step": 1179
+    },
+    {
+      "epoch": 2.617859123682751,
+      "grad_norm": 2.481645345687866,
+      "learning_rate": 4.19753086419753e-05,
+      "loss": 3.35,
+      "step": 1180
+    },
+    {
+      "epoch": 2.620077648363838,
+      "grad_norm": 2.755143642425537,
+      "learning_rate": 4.172839506172839e-05,
+      "loss": 3.3568,
+      "step": 1181
+    },
+    {
+      "epoch": 2.622296173044925,
+      "grad_norm": 2.7017669677734375,
+      "learning_rate": 4.148148148148147e-05,
+      "loss": 3.6146,
+      "step": 1182
+    },
+    {
+      "epoch": 2.624514697726012,
+      "grad_norm": 3.6191043853759766,
+      "learning_rate": 4.123456790123457e-05,
+      "loss": 3.145,
+      "step": 1183
+    },
+    {
+      "epoch": 2.626733222407099,
+      "grad_norm": 2.9682188034057617,
+      "learning_rate": 4.098765432098765e-05,
+      "loss": 3.0971,
+      "step": 1184
+    },
+    {
+      "epoch": 2.6289517470881862,
+      "grad_norm": 2.9087960720062256,
+      "learning_rate": 4.074074074074074e-05,
+      "loss": 3.4838,
+      "step": 1185
+    },
+    {
+      "epoch": 2.6311702717692733,
+      "grad_norm": 3.170659065246582,
+      "learning_rate": 4.0493827160493827e-05,
+      "loss": 3.2321,
+      "step": 1186
+    },
+    {
+      "epoch": 2.6333887964503604,
+      "grad_norm": 2.9661600589752197,
+      "learning_rate": 4.0246913580246915e-05,
+      "loss": 3.6027,
+      "step": 1187
+    },
+    {
+      "epoch": 2.6356073211314475,
+      "grad_norm": 2.8573498725891113,
+      "learning_rate": 3.9999999999999996e-05,
+      "loss": 3.2614,
+      "step": 1188
+    },
+    {
+      "epoch": 2.6378258458125345,
+      "grad_norm": 2.5643155574798584,
+      "learning_rate": 3.975308641975308e-05,
+      "loss": 3.2894,
+      "step": 1189
+    },
+    {
+      "epoch": 2.6400443704936216,
+      "grad_norm": 2.992838144302368,
+      "learning_rate": 3.950617283950617e-05,
+      "loss": 3.4383,
+      "step": 1190
+    },
+    {
+      "epoch": 2.6422628951747087,
+      "grad_norm": 2.8414340019226074,
+      "learning_rate": 3.9259259259259255e-05,
+      "loss": 3.1585,
+      "step": 1191
+    },
+    {
+      "epoch": 2.6444814198557958,
+      "grad_norm": 2.8808953762054443,
+      "learning_rate": 3.901234567901234e-05,
+      "loss": 3.2938,
+      "step": 1192
+    },
+    {
+      "epoch": 2.646699944536883,
+      "grad_norm": 3.060795783996582,
+      "learning_rate": 3.8765432098765425e-05,
+      "loss": 3.6406,
+      "step": 1193
+    },
+    {
+      "epoch": 2.64891846921797,
+      "grad_norm": 2.681269407272339,
+      "learning_rate": 3.851851851851852e-05,
+      "loss": 3.2648,
+      "step": 1194
+    },
+    {
+      "epoch": 2.651136993899057,
+      "grad_norm": 2.659618377685547,
+      "learning_rate": 3.82716049382716e-05,
+      "loss": 3.0524,
+      "step": 1195
+    },
+    {
+      "epoch": 2.653355518580144,
+      "grad_norm": 2.8908562660217285,
+      "learning_rate": 3.802469135802468e-05,
+      "loss": 3.1431,
+      "step": 1196
+    },
+    {
+      "epoch": 2.655574043261231,
+      "grad_norm": 2.7102560997009277,
+      "learning_rate": 3.777777777777777e-05,
+      "loss": 3.0075,
+      "step": 1197
+    },
+    {
+      "epoch": 2.657792567942318,
+      "grad_norm": 2.689999580383301,
+      "learning_rate": 3.7530864197530867e-05,
+      "loss": 3.0997,
+      "step": 1198
+    },
+    {
+      "epoch": 2.6600110926234053,
+      "grad_norm": 2.5635619163513184,
+      "learning_rate": 3.728395061728395e-05,
+      "loss": 3.1437,
+      "step": 1199
+    },
+    {
+      "epoch": 2.6622296173044924,
+      "grad_norm": 3.274951457977295,
+      "learning_rate": 3.703703703703703e-05,
+      "loss": 3.3279,
+      "step": 1200
+    },
+    {
+      "epoch": 2.6644481419855794,
+      "grad_norm": 2.7165257930755615,
+      "learning_rate": 3.679012345679012e-05,
+      "loss": 3.1223,
+      "step": 1201
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 2.8203489780426025,
+      "learning_rate": 3.6543209876543207e-05,
+      "loss": 3.5411,
+      "step": 1202
+    },
+    {
+      "epoch": 2.6688851913477536,
+      "grad_norm": 2.757976770401001,
+      "learning_rate": 3.6296296296296295e-05,
+      "loss": 3.4914,
+      "step": 1203
+    },
+    {
+      "epoch": 2.6711037160288407,
+      "grad_norm": 2.7875313758850098,
+      "learning_rate": 3.604938271604938e-05,
+      "loss": 3.5673,
+      "step": 1204
+    },
+    {
+      "epoch": 2.6733222407099277,
+      "grad_norm": 2.8497684001922607,
+      "learning_rate": 3.5802469135802465e-05,
+      "loss": 3.2683,
+      "step": 1205
+    },
+    {
+      "epoch": 2.675540765391015,
+      "grad_norm": 2.820525884628296,
+      "learning_rate": 3.555555555555555e-05,
+      "loss": 3.1772,
+      "step": 1206
+    },
+    {
+      "epoch": 2.677759290072102,
+      "grad_norm": 3.4860944747924805,
+      "learning_rate": 3.5308641975308635e-05,
+      "loss": 3.1611,
+      "step": 1207
+    },
+    {
+      "epoch": 2.679977814753189,
+      "grad_norm": 2.713298797607422,
+      "learning_rate": 3.506172839506172e-05,
+      "loss": 3.4396,
+      "step": 1208
+    },
+    {
+      "epoch": 2.682196339434276,
+      "grad_norm": 2.86911678314209,
+      "learning_rate": 3.481481481481481e-05,
+      "loss": 3.2754,
+      "step": 1209
+    },
+    {
+      "epoch": 2.684414864115363,
+      "grad_norm": 2.7521750926971436,
+      "learning_rate": 3.45679012345679e-05,
+      "loss": 3.252,
+      "step": 1210
+    },
+    {
+      "epoch": 2.68663338879645,
+      "grad_norm": 2.8987812995910645,
+      "learning_rate": 3.432098765432098e-05,
+      "loss": 3.2163,
+      "step": 1211
+    },
+    {
+      "epoch": 2.6888519134775377,
+      "grad_norm": 2.2600326538085938,
+      "learning_rate": 3.407407407407407e-05,
+      "loss": 3.3847,
+      "step": 1212
+    },
+    {
+      "epoch": 2.691070438158625,
+      "grad_norm": 3.022326946258545,
+      "learning_rate": 3.382716049382716e-05,
+      "loss": 3.6844,
+      "step": 1213
+    },
+    {
+      "epoch": 2.693288962839712,
+      "grad_norm": 2.678861379623413,
+      "learning_rate": 3.3580246913580247e-05,
+      "loss": 3.1356,
+      "step": 1214
+    },
+    {
+      "epoch": 2.695507487520799,
+      "grad_norm": 2.9068548679351807,
+      "learning_rate": 3.333333333333333e-05,
+      "loss": 3.1981,
+      "step": 1215
+    },
+    {
+      "epoch": 2.697726012201886,
+      "grad_norm": 2.5904767513275146,
+      "learning_rate": 3.3086419753086417e-05,
+      "loss": 3.3864,
+      "step": 1216
+    },
+    {
+      "epoch": 2.699944536882973,
+      "grad_norm": 3.185570478439331,
+      "learning_rate": 3.2839506172839505e-05,
+      "loss": 3.0783,
+      "step": 1217
+    },
+    {
+      "epoch": 2.70216306156406,
+      "grad_norm": 2.6785082817077637,
+      "learning_rate": 3.2592592592592586e-05,
+      "loss": 3.3932,
+      "step": 1218
+    },
+    {
+      "epoch": 2.7043815862451472,
+      "grad_norm": 3.3166112899780273,
+      "learning_rate": 3.2345679012345675e-05,
+      "loss": 3.4263,
+      "step": 1219
+    },
+    {
+      "epoch": 2.7066001109262343,
+      "grad_norm": 2.8936643600463867,
+      "learning_rate": 3.209876543209876e-05,
+      "loss": 3.2823,
+      "step": 1220
+    },
+    {
+      "epoch": 2.7088186356073214,
+      "grad_norm": 3.261037588119507,
+      "learning_rate": 3.185185185185185e-05,
+      "loss": 3.3133,
+      "step": 1221
+    },
+    {
+      "epoch": 2.7110371602884085,
+      "grad_norm": 2.801356077194214,
+      "learning_rate": 3.160493827160493e-05,
+      "loss": 3.2938,
+      "step": 1222
+    },
+    {
+      "epoch": 2.7132556849694955,
+      "grad_norm": 2.574474334716797,
+      "learning_rate": 3.135802469135802e-05,
+      "loss": 3.238,
+      "step": 1223
+    },
+    {
+      "epoch": 2.7154742096505826,
+      "grad_norm": 2.838669538497925,
+      "learning_rate": 3.111111111111111e-05,
+      "loss": 3.4172,
+      "step": 1224
+    },
+    {
+      "epoch": 2.7176927343316697,
+      "grad_norm": 2.9818763732910156,
+      "learning_rate": 3.08641975308642e-05,
+      "loss": 3.292,
+      "step": 1225
+    },
+    {
+      "epoch": 2.7199112590127568,
+      "grad_norm": 2.8964338302612305,
+      "learning_rate": 3.061728395061728e-05,
+      "loss": 3.0666,
+      "step": 1226
+    },
+    {
+      "epoch": 2.722129783693844,
+      "grad_norm": 2.6595730781555176,
+      "learning_rate": 3.0370370370370368e-05,
+      "loss": 3.0834,
+      "step": 1227
+    },
+    {
+      "epoch": 2.724348308374931,
+      "grad_norm": 3.08070707321167,
+      "learning_rate": 3.0123456790123453e-05,
+      "loss": 3.5488,
+      "step": 1228
+    },
+    {
+      "epoch": 2.726566833056018,
+      "grad_norm": 3.102290630340576,
+      "learning_rate": 2.987654320987654e-05,
+      "loss": 3.3146,
+      "step": 1229
+    },
+    {
+      "epoch": 2.728785357737105,
+      "grad_norm": 3.2188045978546143,
+      "learning_rate": 2.9629629629629627e-05,
+      "loss": 3.2729,
+      "step": 1230
+    },
+    {
+      "epoch": 2.731003882418192,
+      "grad_norm": 2.505281925201416,
+      "learning_rate": 2.9382716049382715e-05,
+      "loss": 3.5704,
+      "step": 1231
+    },
+    {
+      "epoch": 2.733222407099279,
+      "grad_norm": 3.2931861877441406,
+      "learning_rate": 2.9135802469135796e-05,
+      "loss": 3.1628,
+      "step": 1232
+    },
+    {
+      "epoch": 2.7354409317803663,
+      "grad_norm": 3.1521599292755127,
+      "learning_rate": 2.8888888888888885e-05,
+      "loss": 2.9257,
+      "step": 1233
+    },
+    {
+      "epoch": 2.7376594564614534,
+      "grad_norm": 2.8271191120147705,
+      "learning_rate": 2.8641975308641973e-05,
+      "loss": 3.363,
+      "step": 1234
+    },
+    {
+      "epoch": 2.7398779811425404,
+      "grad_norm": 2.6765851974487305,
+      "learning_rate": 2.8395061728395058e-05,
+      "loss": 3.5697,
+      "step": 1235
+    },
+    {
+      "epoch": 2.7420965058236275,
+      "grad_norm": 2.8013389110565186,
+      "learning_rate": 2.8148148148148147e-05,
+      "loss": 3.0744,
+      "step": 1236
+    },
+    {
+      "epoch": 2.7443150305047146,
+      "grad_norm": 2.771820545196533,
+      "learning_rate": 2.790123456790123e-05,
+      "loss": 3.3841,
+      "step": 1237
+    },
+    {
+      "epoch": 2.7465335551858017,
+      "grad_norm": 2.808199405670166,
+      "learning_rate": 2.765432098765432e-05,
+      "loss": 3.1057,
+      "step": 1238
+    },
+    {
+      "epoch": 2.7487520798668887,
+      "grad_norm": 3.365356922149658,
+      "learning_rate": 2.7407407407407405e-05,
+      "loss": 3.1684,
+      "step": 1239
+    },
+    {
+      "epoch": 2.750970604547976,
+      "grad_norm": 3.0335288047790527,
+      "learning_rate": 2.7160493827160493e-05,
+      "loss": 3.1153,
+      "step": 1240
+    },
+    {
+      "epoch": 2.753189129229063,
+      "grad_norm": 2.56756591796875,
+      "learning_rate": 2.6913580246913578e-05,
+      "loss": 3.3044,
+      "step": 1241
+    },
+    {
+      "epoch": 2.75540765391015,
+      "grad_norm": 2.8652429580688477,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 3.5829,
+      "step": 1242
+    },
+    {
+      "epoch": 2.757626178591237,
+      "grad_norm": 2.7513015270233154,
+      "learning_rate": 2.6419753086419748e-05,
+      "loss": 3.2784,
+      "step": 1243
+    },
+    {
+      "epoch": 2.759844703272324,
+      "grad_norm": 2.514538526535034,
+      "learning_rate": 2.6172839506172837e-05,
+      "loss": 3.2794,
+      "step": 1244
+    },
+    {
+      "epoch": 2.762063227953411,
+      "grad_norm": 2.913966178894043,
+      "learning_rate": 2.592592592592592e-05,
+      "loss": 3.448,
+      "step": 1245
+    },
+    {
+      "epoch": 2.7642817526344983,
+      "grad_norm": 2.902489423751831,
+      "learning_rate": 2.567901234567901e-05,
+      "loss": 3.346,
+      "step": 1246
+    },
+    {
+      "epoch": 2.7665002773155853,
+      "grad_norm": 2.7837393283843994,
+      "learning_rate": 2.5432098765432095e-05,
+      "loss": 3.3902,
+      "step": 1247
+    },
+    {
+      "epoch": 2.7687188019966724,
+      "grad_norm": 2.490365505218506,
+      "learning_rate": 2.5185185185185183e-05,
+      "loss": 2.9982,
+      "step": 1248
+    },
+    {
+      "epoch": 2.7709373266777595,
+      "grad_norm": 2.6419758796691895,
+      "learning_rate": 2.4938271604938268e-05,
+      "loss": 3.1773,
+      "step": 1249
+    },
+    {
+      "epoch": 2.7731558513588466,
+      "grad_norm": 2.650132179260254,
+      "learning_rate": 2.4691358024691357e-05,
+      "loss": 3.249,
+      "step": 1250
+    },
+    {
+      "epoch": 2.7753743760399336,
+      "grad_norm": 2.8140628337860107,
+      "learning_rate": 2.4444444444444445e-05,
+      "loss": 3.5192,
+      "step": 1251
+    },
+    {
+      "epoch": 2.7775929007210207,
+      "grad_norm": 2.9799649715423584,
+      "learning_rate": 2.419753086419753e-05,
+      "loss": 3.0067,
+      "step": 1252
+    },
+    {
+      "epoch": 2.779811425402108,
+      "grad_norm": 3.091761350631714,
+      "learning_rate": 2.3950617283950618e-05,
+      "loss": 3.4755,
+      "step": 1253
+    },
+    {
+      "epoch": 2.782029950083195,
+      "grad_norm": 2.87955641746521,
+      "learning_rate": 2.37037037037037e-05,
+      "loss": 3.3065,
+      "step": 1254
+    },
+    {
+      "epoch": 2.784248474764282,
+      "grad_norm": 2.434142827987671,
+      "learning_rate": 2.3456790123456788e-05,
+      "loss": 3.571,
+      "step": 1255
+    },
+    {
+      "epoch": 2.786466999445369,
+      "grad_norm": 2.523686170578003,
+      "learning_rate": 2.3209876543209873e-05,
+      "loss": 3.128,
+      "step": 1256
+    },
+    {
+      "epoch": 2.788685524126456,
+      "grad_norm": 3.0034890174865723,
+      "learning_rate": 2.296296296296296e-05,
+      "loss": 3.5448,
+      "step": 1257
+    },
+    {
+      "epoch": 2.790904048807543,
+      "grad_norm": 2.4801077842712402,
+      "learning_rate": 2.2716049382716047e-05,
+      "loss": 3.0517,
+      "step": 1258
+    },
+    {
+      "epoch": 2.7931225734886302,
+      "grad_norm": 2.8369855880737305,
+      "learning_rate": 2.2469135802469135e-05,
+      "loss": 3.431,
+      "step": 1259
+    },
+    {
+      "epoch": 2.7953410981697173,
+      "grad_norm": 3.0403037071228027,
+      "learning_rate": 2.222222222222222e-05,
+      "loss": 3.3735,
+      "step": 1260
+    },
+    {
+      "epoch": 2.7975596228508044,
+      "grad_norm": 3.029148578643799,
+      "learning_rate": 2.1975308641975308e-05,
+      "loss": 2.9732,
+      "step": 1261
+    },
+    {
+      "epoch": 2.7997781475318915,
+      "grad_norm": 2.5942745208740234,
+      "learning_rate": 2.1728395061728393e-05,
+      "loss": 3.225,
+      "step": 1262
+    },
+    {
+      "epoch": 2.8019966722129785,
+      "grad_norm": 2.7814176082611084,
+      "learning_rate": 2.148148148148148e-05,
+      "loss": 3.1927,
+      "step": 1263
+    },
+    {
+      "epoch": 2.8042151968940656,
+      "grad_norm": 2.654717445373535,
+      "learning_rate": 2.1234567901234563e-05,
+      "loss": 3.1839,
+      "step": 1264
+    },
+    {
+      "epoch": 2.8064337215751527,
+      "grad_norm": 3.195948839187622,
+      "learning_rate": 2.098765432098765e-05,
+      "loss": 3.4836,
+      "step": 1265
+    },
+    {
+      "epoch": 2.8086522462562398,
+      "grad_norm": 2.8370094299316406,
+      "learning_rate": 2.0740740740740737e-05,
+      "loss": 3.1494,
+      "step": 1266
+    },
+    {
+      "epoch": 2.810870770937327,
+      "grad_norm": 2.658625364303589,
+      "learning_rate": 2.0493827160493825e-05,
+      "loss": 3.2559,
+      "step": 1267
+    },
+    {
+      "epoch": 2.813089295618414,
+      "grad_norm": 2.878044843673706,
+      "learning_rate": 2.0246913580246913e-05,
+      "loss": 3.2208,
+      "step": 1268
+    },
+    {
+      "epoch": 2.815307820299501,
+      "grad_norm": 2.314476251602173,
+      "learning_rate": 1.9999999999999998e-05,
+      "loss": 3.3171,
+      "step": 1269
+    },
+    {
+      "epoch": 2.817526344980588,
+      "grad_norm": 2.3043107986450195,
+      "learning_rate": 1.9753086419753087e-05,
+      "loss": 3.1833,
+      "step": 1270
+    },
+    {
+      "epoch": 2.819744869661675,
+      "grad_norm": 2.773775577545166,
+      "learning_rate": 1.950617283950617e-05,
+      "loss": 3.2744,
+      "step": 1271
+    },
+    {
+      "epoch": 2.821963394342762,
+      "grad_norm": 2.5326461791992188,
+      "learning_rate": 1.925925925925926e-05,
+      "loss": 3.0924,
+      "step": 1272
+    },
+    {
+      "epoch": 2.8241819190238493,
+      "grad_norm": 2.7197911739349365,
+      "learning_rate": 1.901234567901234e-05,
+      "loss": 3.177,
+      "step": 1273
+    },
+    {
+      "epoch": 2.8264004437049364,
+      "grad_norm": 2.445826768875122,
+      "learning_rate": 1.8765432098765433e-05,
+      "loss": 3.079,
+      "step": 1274
+    },
+    {
+      "epoch": 2.8286189683860234,
+      "grad_norm": 2.83439564704895,
+      "learning_rate": 1.8518518518518515e-05,
+      "loss": 3.2904,
+      "step": 1275
+    },
+    {
+      "epoch": 2.8308374930671105,
+      "grad_norm": 2.8362514972686768,
+      "learning_rate": 1.8271604938271603e-05,
+      "loss": 3.1347,
+      "step": 1276
+    },
+    {
+      "epoch": 2.8330560177481976,
+      "grad_norm": 2.550525188446045,
+      "learning_rate": 1.802469135802469e-05,
+      "loss": 3.2839,
+      "step": 1277
+    },
+    {
+      "epoch": 2.8352745424292847,
+      "grad_norm": 3.1011059284210205,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 3.4093,
+      "step": 1278
+    },
+    {
+      "epoch": 2.8374930671103717,
+      "grad_norm": 2.670053482055664,
+      "learning_rate": 1.753086419753086e-05,
+      "loss": 3.1897,
+      "step": 1279
+    },
+    {
+      "epoch": 2.839711591791459,
+      "grad_norm": 3.248617649078369,
+      "learning_rate": 1.728395061728395e-05,
+      "loss": 3.3919,
+      "step": 1280
+    },
+    {
+      "epoch": 2.841930116472546,
+      "grad_norm": 3.11806321144104,
+      "learning_rate": 1.7037037037037035e-05,
+      "loss": 3.0373,
+      "step": 1281
+    },
+    {
+      "epoch": 2.844148641153633,
+      "grad_norm": 2.654214859008789,
+      "learning_rate": 1.6790123456790123e-05,
+      "loss": 3.579,
+      "step": 1282
+    },
+    {
+      "epoch": 2.84636716583472,
+      "grad_norm": 3.104462146759033,
+      "learning_rate": 1.6543209876543208e-05,
+      "loss": 3.1468,
+      "step": 1283
+    },
+    {
+      "epoch": 2.848585690515807,
+      "grad_norm": 2.9722726345062256,
+      "learning_rate": 1.6296296296296293e-05,
+      "loss": 3.3317,
+      "step": 1284
+    },
+    {
+      "epoch": 2.850804215196894,
+      "grad_norm": 2.548308849334717,
+      "learning_rate": 1.604938271604938e-05,
+      "loss": 2.9777,
+      "step": 1285
+    },
+    {
+      "epoch": 2.8530227398779813,
+      "grad_norm": 2.8028459548950195,
+      "learning_rate": 1.5802469135802467e-05,
+      "loss": 3.105,
+      "step": 1286
+    },
+    {
+      "epoch": 2.8552412645590683,
+      "grad_norm": 2.7234795093536377,
+      "learning_rate": 1.5555555555555555e-05,
+      "loss": 2.6936,
+      "step": 1287
+    },
+    {
+      "epoch": 2.8574597892401554,
+      "grad_norm": 2.5337350368499756,
+      "learning_rate": 1.530864197530864e-05,
+      "loss": 3.3851,
+      "step": 1288
+    },
+    {
+      "epoch": 2.8596783139212425,
+      "grad_norm": 3.903007745742798,
+      "learning_rate": 1.5061728395061727e-05,
+      "loss": 3.1174,
+      "step": 1289
+    },
+    {
+      "epoch": 2.8618968386023296,
+      "grad_norm": 2.599226236343384,
+      "learning_rate": 1.4814814814814813e-05,
+      "loss": 3.235,
+      "step": 1290
+    },
+    {
+      "epoch": 2.8641153632834166,
+      "grad_norm": 2.820854425430298,
+      "learning_rate": 1.4567901234567898e-05,
+      "loss": 3.1985,
+      "step": 1291
+    },
+    {
+      "epoch": 2.8663338879645037,
+      "grad_norm": 3.24422550201416,
+      "learning_rate": 1.4320987654320987e-05,
+      "loss": 3.0765,
+      "step": 1292
+    },
+    {
+      "epoch": 2.868552412645591,
+      "grad_norm": 2.713667154312134,
+      "learning_rate": 1.4074074074074073e-05,
+      "loss": 3.2768,
+      "step": 1293
+    },
+    {
+      "epoch": 2.870770937326678,
+      "grad_norm": 2.44154691696167,
+      "learning_rate": 1.382716049382716e-05,
+      "loss": 3.3041,
+      "step": 1294
+    },
+    {
+      "epoch": 2.872989462007765,
+      "grad_norm": 3.180100202560425,
+      "learning_rate": 1.3580246913580247e-05,
+      "loss": 3.3566,
+      "step": 1295
+    },
+    {
+      "epoch": 2.875207986688852,
+      "grad_norm": 2.4062674045562744,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 3.2539,
+      "step": 1296
+    },
+    {
+      "epoch": 2.877426511369939,
+      "grad_norm": 2.8133068084716797,
+      "learning_rate": 1.3086419753086418e-05,
+      "loss": 3.0213,
+      "step": 1297
+    },
+    {
+      "epoch": 2.879645036051026,
+      "grad_norm": 3.317291021347046,
+      "learning_rate": 1.2839506172839505e-05,
+      "loss": 3.3148,
+      "step": 1298
+    },
+    {
+      "epoch": 2.8818635607321132,
+      "grad_norm": 2.510012149810791,
+      "learning_rate": 1.2592592592592592e-05,
+      "loss": 3.3949,
+      "step": 1299
+    },
+    {
+      "epoch": 2.8840820854132003,
+      "grad_norm": 2.6490437984466553,
+      "learning_rate": 1.2345679012345678e-05,
+      "loss": 3.1794,
+      "step": 1300
+    },
+    {
+      "epoch": 2.8863006100942874,
+      "grad_norm": 2.6988909244537354,
+      "learning_rate": 1.2098765432098765e-05,
+      "loss": 3.0529,
+      "step": 1301
+    },
+    {
+      "epoch": 2.8885191347753745,
+      "grad_norm": 2.6287052631378174,
+      "learning_rate": 1.185185185185185e-05,
+      "loss": 3.1722,
+      "step": 1302
+    },
+    {
+      "epoch": 2.8907376594564616,
+      "grad_norm": 2.9318556785583496,
+      "learning_rate": 1.1604938271604937e-05,
+      "loss": 3.0106,
+      "step": 1303
+    },
+    {
+      "epoch": 2.8929561841375486,
+      "grad_norm": 2.6143076419830322,
+      "learning_rate": 1.1358024691358023e-05,
+      "loss": 3.2133,
+      "step": 1304
+    },
+    {
+      "epoch": 2.8951747088186357,
+      "grad_norm": 2.8240926265716553,
+      "learning_rate": 1.111111111111111e-05,
+      "loss": 3.1177,
+      "step": 1305
+    },
+    {
+      "epoch": 2.8973932334997228,
+      "grad_norm": 2.538440465927124,
+      "learning_rate": 1.0864197530864197e-05,
+      "loss": 3.2356,
+      "step": 1306
+    },
+    {
+      "epoch": 2.89961175818081,
+      "grad_norm": 3.4087648391723633,
+      "learning_rate": 1.0617283950617282e-05,
+      "loss": 3.1963,
+      "step": 1307
+    },
+    {
+      "epoch": 2.901830282861897,
+      "grad_norm": 2.337791681289673,
+      "learning_rate": 1.0370370370370368e-05,
+      "loss": 3.3164,
+      "step": 1308
+    },
+    {
+      "epoch": 2.904048807542984,
+      "grad_norm": 2.7128920555114746,
+      "learning_rate": 1.0123456790123457e-05,
+      "loss": 3.0935,
+      "step": 1309
+    },
+    {
+      "epoch": 2.906267332224071,
+      "grad_norm": 3.1189403533935547,
+      "learning_rate": 9.876543209876543e-06,
+      "loss": 3.0705,
+      "step": 1310
+    },
+    {
+      "epoch": 2.908485856905158,
+      "grad_norm": 2.4013233184814453,
+      "learning_rate": 9.62962962962963e-06,
+      "loss": 3.2107,
+      "step": 1311
+    },
+    {
+      "epoch": 2.9107043815862452,
+      "grad_norm": 2.628100633621216,
+      "learning_rate": 9.382716049382717e-06,
+      "loss": 3.1514,
+      "step": 1312
+    },
+    {
+      "epoch": 2.9129229062673323,
+      "grad_norm": 2.9449737071990967,
+      "learning_rate": 9.135802469135802e-06,
+      "loss": 2.9522,
+      "step": 1313
+    },
+    {
+      "epoch": 2.9151414309484194,
+      "grad_norm": 2.9202868938446045,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 3.2157,
+      "step": 1314
+    },
+    {
+      "epoch": 2.9173599556295065,
+      "grad_norm": 3.178095579147339,
+      "learning_rate": 8.641975308641975e-06,
+      "loss": 3.2257,
+      "step": 1315
+    },
+    {
+      "epoch": 2.9195784803105935,
+      "grad_norm": 2.4233884811401367,
+      "learning_rate": 8.395061728395062e-06,
+      "loss": 3.4881,
+      "step": 1316
+    },
+    {
+      "epoch": 2.9217970049916806,
+      "grad_norm": 2.4404220581054688,
+      "learning_rate": 8.148148148148147e-06,
+      "loss": 3.0759,
+      "step": 1317
+    },
+    {
+      "epoch": 2.9240155296727677,
+      "grad_norm": 2.8211722373962402,
+      "learning_rate": 7.901234567901233e-06,
+      "loss": 3.4396,
+      "step": 1318
+    },
+    {
+      "epoch": 2.9262340543538548,
+      "grad_norm": 2.604445457458496,
+      "learning_rate": 7.65432098765432e-06,
+      "loss": 3.3796,
+      "step": 1319
+    },
+    {
+      "epoch": 2.928452579034942,
+      "grad_norm": 2.5345306396484375,
+      "learning_rate": 7.407407407407407e-06,
+      "loss": 3.4037,
+      "step": 1320
+    },
+    {
+      "epoch": 2.930671103716029,
+      "grad_norm": 2.658292531967163,
+      "learning_rate": 7.160493827160493e-06,
+      "loss": 3.1939,
+      "step": 1321
+    },
+    {
+      "epoch": 2.932889628397116,
+      "grad_norm": 3.0478193759918213,
+      "learning_rate": 6.91358024691358e-06,
+      "loss": 3.5897,
+      "step": 1322
+    },
+    {
+      "epoch": 2.935108153078203,
+      "grad_norm": 2.871546506881714,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 3.2137,
+      "step": 1323
+    },
+    {
+      "epoch": 2.93732667775929,
+      "grad_norm": 2.680933952331543,
+      "learning_rate": 6.4197530864197525e-06,
+      "loss": 3.1781,
+      "step": 1324
+    },
+    {
+      "epoch": 2.939545202440377,
+      "grad_norm": 2.5698466300964355,
+      "learning_rate": 6.172839506172839e-06,
+      "loss": 3.1628,
+      "step": 1325
+    },
+    {
+      "epoch": 2.9417637271214643,
+      "grad_norm": 2.3731906414031982,
+      "learning_rate": 5.925925925925925e-06,
+      "loss": 3.2585,
+      "step": 1326
+    },
+    {
+      "epoch": 2.9439822518025514,
+      "grad_norm": 2.811649799346924,
+      "learning_rate": 5.679012345679012e-06,
+      "loss": 3.2756,
+      "step": 1327
+    },
+    {
+      "epoch": 2.9462007764836384,
+      "grad_norm": 2.662019729614258,
+      "learning_rate": 5.432098765432098e-06,
+      "loss": 3.4567,
+      "step": 1328
+    },
+    {
+      "epoch": 2.9484193011647255,
+      "grad_norm": 2.7594850063323975,
+      "learning_rate": 5.185185185185184e-06,
+      "loss": 3.0653,
+      "step": 1329
+    },
+    {
+      "epoch": 2.9506378258458126,
+      "grad_norm": 2.850553035736084,
+      "learning_rate": 4.938271604938272e-06,
+      "loss": 3.3359,
+      "step": 1330
+    },
+    {
+      "epoch": 2.9528563505268997,
+      "grad_norm": 2.5800914764404297,
+      "learning_rate": 4.691358024691358e-06,
+      "loss": 3.5194,
+      "step": 1331
+    },
+    {
+      "epoch": 2.9550748752079867,
+      "grad_norm": 2.810091972351074,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 3.4085,
+      "step": 1332
+    },
+    {
+      "epoch": 2.957293399889074,
+      "grad_norm": 3.0693106651306152,
+      "learning_rate": 4.197530864197531e-06,
+      "loss": 3.1082,
+      "step": 1333
+    },
+    {
+      "epoch": 2.959511924570161,
+      "grad_norm": 2.707022190093994,
+      "learning_rate": 3.950617283950617e-06,
+      "loss": 3.3872,
+      "step": 1334
+    },
+    {
+      "epoch": 2.961730449251248,
+      "grad_norm": 2.499107837677002,
+      "learning_rate": 3.7037037037037033e-06,
+      "loss": 3.2625,
+      "step": 1335
+    },
+    {
+      "epoch": 2.963948973932335,
+      "grad_norm": 2.7464599609375,
+      "learning_rate": 3.45679012345679e-06,
+      "loss": 3.2268,
+      "step": 1336
+    },
+    {
+      "epoch": 2.966167498613422,
+      "grad_norm": 2.5510807037353516,
+      "learning_rate": 3.2098765432098762e-06,
+      "loss": 3.1646,
+      "step": 1337
+    },
+    {
+      "epoch": 2.968386023294509,
+      "grad_norm": 2.8641560077667236,
+      "learning_rate": 2.9629629629629625e-06,
+      "loss": 3.3725,
+      "step": 1338
+    },
+    {
+      "epoch": 2.9706045479755963,
+      "grad_norm": 3.037064552307129,
+      "learning_rate": 2.716049382716049e-06,
+      "loss": 3.2013,
+      "step": 1339
+    },
+    {
+      "epoch": 2.9728230726566833,
+      "grad_norm": 3.046401023864746,
+      "learning_rate": 2.469135802469136e-06,
+      "loss": 3.4146,
+      "step": 1340
+    },
+    {
+      "epoch": 2.9750415973377704,
+      "grad_norm": 2.7076127529144287,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 3.1586,
+      "step": 1341
+    },
+    {
+      "epoch": 2.9772601220188575,
+      "grad_norm": 2.4789509773254395,
+      "learning_rate": 1.9753086419753083e-06,
+      "loss": 3.2659,
+      "step": 1342
+    },
+    {
+      "epoch": 2.9794786466999446,
+      "grad_norm": 2.754330635070801,
+      "learning_rate": 1.728395061728395e-06,
+      "loss": 3.1956,
+      "step": 1343
+    },
+    {
+      "epoch": 2.9816971713810316,
+      "grad_norm": 2.8958003520965576,
+      "learning_rate": 1.4814814814814812e-06,
+      "loss": 3.3205,
+      "step": 1344
+    },
+    {
+      "epoch": 2.9839156960621187,
+      "grad_norm": 2.3046891689300537,
+      "learning_rate": 1.234567901234568e-06,
+      "loss": 3.5233,
+      "step": 1345
+    },
+    {
+      "epoch": 2.986134220743206,
+      "grad_norm": 3.1731839179992676,
+      "learning_rate": 9.876543209876542e-07,
+      "loss": 3.4537,
+      "step": 1346
+    },
+    {
+      "epoch": 2.988352745424293,
+      "grad_norm": 2.366079568862915,
+      "learning_rate": 7.407407407407406e-07,
+      "loss": 3.0703,
+      "step": 1347
+    },
+    {
+      "epoch": 2.99057127010538,
+      "grad_norm": 2.592994213104248,
+      "learning_rate": 4.938271604938271e-07,
+      "loss": 3.3386,
+      "step": 1348
+    },
+    {
+      "epoch": 2.992789794786467,
+      "grad_norm": 2.8718955516815186,
+      "learning_rate": 2.4691358024691354e-07,
+      "loss": 3.2448,
+      "step": 1349
+    },
+    {
+      "epoch": 2.995008319467554,
+      "grad_norm": 2.9335429668426514,
+      "learning_rate": 0.0,
+      "loss": 3.174,
+      "step": 1350
+    },
+    {
+      "epoch": 2.995008319467554,
+      "step": 1350,
+      "total_flos": 6.970992786879283e+16,
+      "train_loss": 3.9628717678564565,
+      "train_runtime": 1531.0104,
+      "train_samples_per_second": 14.13,
+      "train_steps_per_second": 0.882
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1350,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.970992786879283e+16,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}